Professional Documents
Culture Documents
This manuscript has been reproduced from the microfilm m aster. UMI films
the text directly from the original or copy submitted. Thus, som e th esis and
dissertation copies are in typewriter face, while others may t>e from any type of
computer printer.
In the unlikely event that the author did not send UMI a complete manuscript
and there are missing pages, these will t>e noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
UMI'
NOTE TO USERS
UMI
UNIVERSITY OF OKLAHOMA
GRADUATE COLLEGE
A DISSERTATION
degree of
By
SHUNCHENG J I
Norman, Oklahoma
2001
UMI N um ber: 3029622
UMI
UMI Microform 3029622
Copyright 2002 by Bell & Howell Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
A DISSERTATION
By:
Misnel
aruk Civan
Lance Lobban
/Zcj
Pakiz&Pulat
Daniel ResSsco
c Copyright by Shuncheng Ji 2001
All Rights Reserved.
ACKNOW LEDGEM ENTS
I would like to thank all o f the Chemical Engineering Faculty and Staff at the
University of Oklahoma for their support and friendship, especially Dr. Richard
Mallinson, who provided me partial support for one and a h alf years, and Mr. Rick
I would like to thank Dr. Faruk Civan, Dr. Lance Lobban, Dr. Pakize Pulat and
Dr. Daniel Resasco for their participation on my dissertation committee. I will be very
grateful to Dr. Bagajewicz for his guidance and patience through the entire dissertation
work.
IV
TABLE OF CONTENTS
Chapter 1
Introduction...................................................................................................................................1
1.6 References............................................................................................................................ 18
2.1 Overview.............................................................................................................................. 20
2.2 Introduction..........................................................................................................................20
2.6 Illustration...........................................................................................................................35
2.8 Flexibility............................................................................................................................55
2.9 Reboiler............................................................................................................................... 55
2.10 Applicability to other crudes............................................................................................57
2.11 Conclusion......................................................................................................................... 58
2.12 Nomenclature....................................................................................................................58
2.13 References.......................................................................................................................... 60
Chapter 3 Rigorous Targeting Procedure for the Design o f Crude Fractionation with Pre
flashing........................................................................................................................................ 63
3.1 Overview...............................................................................................................................63
3.2 Introduction.......................................................................................................................... 63
3.12 Conclusion......................................................................................................................... 95
3.13 Nomenclature.....................................................................................................................96
3.14 References.......................................................................................................................... 97
4.1 Overview.............................................................................................................................. 99
4.2 Introduction.......................................................................................................................... 99
VI
4.3 Crude vaporization patterns and heat demand..............................................................103
4.7 Conclusions.......................................................................................................................121
4.8 References........................................................................................................................114
4.9 Appendix..........................................................................................................................122
Chapter 5 Design o f Crude Distillation Plants with Vacuum Units. I. Targeting........ 126
5.1 Overview..........................................................................................................................126
5.9 Conclusion........................................................................................................................154
5.11 Appendix........................................................................................................................158
Chapter 6 Design o f Crude Distillation Plants with Vacuum Units. II. HEN Design.... 161
6.1 Overview...........................................................................................................................161
VII
6.4 Heat exchanger network.................................................................................................. 170
7.5 References.........................................................................................................................204
vin
ABSTRACT
Crude distillation is the first processing unit in the refinery. It separates the crude
in to distillates, which are either intermediate products further processed or raw materials
for the chemical industry. Crude distillation consumes fuel in the equivalent of 2% o f the
crude processed. This dissertation proposes a rigorous procedure that uses simulators that
are able to rigorously represent the column operating conditions as a function o f the
design parameters and complements it with heat exchanger network design methods.
Targeting procedures for both atmospheric distillations and complete distillations are
presented. These procedures use a heat demand-supply diagram and commercial software
to determine the energy targets for a plant processing several crude oils. These
alternative designs, the preflash design and stripping-type design have been studied in
detail. A rigorous definition for “carrier effect” is provided and the role of light
components in the crude distillation is analyzed. The impact o f the different heating
pattern in the stripping-type design is analyzed and compared with the conventional
design. It is concluded that that a stripping-type distillation is less competitive than the
conventional design in terms o f energy consumption and investment cost. The problem o f
addressed. The plant has a preflash drum to facilitate the processing of light crudes. A
new concept, matching pattern, is proposed to capture the heat exchange topology
features. The dissertation also proposes a methodology to develop new designs and
IX
Chapter 1 Introduction
Crude distillation is the first processing unit in the refinery. It separates the crude
in to distillates, which are either intermediate products further processed or raw materials
for the chemical industry. Crude distillation consumes fuel in the equivalent of 2% o f the
crude processed. Because o f its large scale, even a small percentage of energy reduction
is economically significant. In spite of its importance, procedures for its optimal design
A schematic diagram for crude distillation is shown in Figure 1.1. The crude oil is
split in the atmospheric tower into fuel gas, naphtha, kerosene, diesel, gas oil and
atmospheric residue (topped crude). The atmospheric residue is sent to the vacuum tower
to be further separated into light vacuum gas oil, heavy vacuum gas oil and vacuum
residue.
naphtha To Ejectors
Preflash drum
steam
kerosene
VLGO
steam
Crude Oil
diesel VHGO
Steam
Vacuum Residue
sour water Furnace
Vacuum furnace
Desalter
a simplified version o f the atmospheric column. There are five products coming out from
this column: naphtha from the top, residue from the bottom, kerosene, diesel and gas oil
from the intermediate trays. The three products (kerosene, diesel and gas oil) withdrawn
from the side o f the column are also called side withdrawals.
The side withdrawals can not meet their specifications because they contain
substantial amount o f light components. The light components are removed in side
» water
^ naphtha
kerosene
diesel
Flash zone
steam
Residue
steam Stripper 3
►kerosene
steam Stripper 2
diesel
Residue
Figure 1.3 The atmospheric distillation column with side strippers (design B)
Design B is a feasible scheme but not a good one. In this design, all heat is
removed from the condenser. This results in vapor and liquid traffic increasing
throughout the tower from the bottom to the top. The diameter o f the top section o f the
column has to be large to accommodate the high traffic (Watkins, 1979). This scheme of
heat removal is also not appropriate for heat recovery because heat from the condenser at
Design B can be improved by removing part o f the heat somewhere below the top
tray. There are two types o f schemes (Watkins, 1979): pump-around reflux and pump-
back reflux.
The pump-back scheme (Figure 1.4) withdraws the liquid from a tray, cool it
down and pumps it back to the column at a point several trays below the withdrawal
point. Note the amount o f heat that can be removed by the pump-back scheme is limited
water
►naphtha
Pump-back 1
‘A
steam Stripper 3
Pump-back 2
C kerosene
steam Stripper 2
Pump-back3 ^ diesel
gas oil
Flash zone
steam
Residue
The pump-around scheme (Figure 1.5) uses an externally circulated and cooled
stream for partial heat removal. Because the liquid is recycled in the pump-around zone,
the amount of liquid withdrawn can be several times larger than the liquid reflux entering
the pump-around zone. The pump-around scheme is the design used in the refineries.
water
naphtha
PAl
c :
steam Stripper 3
P A 2 ^
kerosene
H j steam Stripper 2
PA3
c :
Crude steam Stripper 1
gas oil
Flash zone
steam
Residue
Figure 1.5 Pump-around reflux — some heat removed from the pump-around
circuits
Thus, the conventional design using side strippers and pump-arounds (Figure 1.5)
appeared 70 years ago and still the dominant design. Because the trade-off between
energy cost and capital cost has changed profoundly in the past 70 years, it is necessary
to check whether the assumptions used to design this process are still valid and propose
an updated procedure. Existing procedures (Watkins, 1979) are not accurate in nature
because they rely on empirical charts. Another problem is that these procedures do not
take into account the impact o f heat integration and are therefore energy inefficient. In
addition, new methodologies are available for tackling certain design problems that did
Another motivation for this thesis is to study some interesting design alternatives.
(Figure 1.6) is practiced sometimes. There have always been disputes as to whether these
designs could lead to smaller energy consumption. This thesis evaluates these designs in
a systematic way.
i condenser
-►
desalter — » r *1 Steam
stean
C=l PA3
c : diesel
sour water
A Steam
gas oil
• - 0 - steam
fumace ► residue
prefractionation
tower mam tower
Figure 1.6 Pre-fractionation Design
determine the distillation column(s) and the associated heat exchanger network that will
minimize the annualized cost o f the equipment plus the annual cost o f utilities.
Crude oil is very complex. The compounds in crude oils can be grouped into 3
been identified from very light hydrocarbons (less than 10 carbon atoms) to very heavy
large hydrocarbon molecules. The relevant properties to crude distillation are boiling
ranges, density and salt content. Boiling ranges are used because of the large amount of
compounds which are very expensive to identify. Density helps characterize the crude
because it tells about its proportion of light hydrocarbons. Finally salt comes with the
The boiling range of the crude oil provides information about the quantities of the
various products present. The commonly used distillations are true boiling point (TBP)
True Boiling Point (TBP) distillation is a batch distillation using a large number
of trays and a high reflux ratio. The temperature-volumetric yield curve is constructed
representing the actual boiling point o f the hydrocarbon material present at the volume
percentage point (Watkins, 1979). Most commonly the column has 15 theoretical trays
and is operated at a reflux ratio o f 5. TBP distillations are usually reported at 760 mm Hg,
even though the highest boiling portion of the mixture is distilled out at reduced pressures
(30 to 40 mm Hg). The TBP data (Figure 1.7) can be used to define the average normal
boiling points for pseudocomponents. The pseudocomponent concept is very useful for
simulations. During the TBP distillation, if the distillates are collected over a small
temperature range, the whole crude is then divided in fractions. Each fraction is still a
mixture o f m any unknown components, but the components in one fraction have a similar
boiling point. It would be convenient for distillation calculations to treat these fractions as
distillation curve. Figure 1.8 shows the first three pseudocomponents. It can be seen that
a psedocomponent is defined by the interval between two adjacent cut points and its
percentage in the crude is the differences between values at adjacent cut points. The
1000
400
LÜ
& 200
S
UJ
I-
-200
0 20 40 60 80 100
PERCENT DISTILLED
1 00
80
60
U
UJ
ae 40
I
UJ
20
Q.
S
-20
-40 2nd 3rd P C
PC PC
-60
0 2 4 6 8 10 12
P E R C E N T DISTILLED
PC: pseudo-component
The disadvantage o f TBP distillation is that the equipment is expensive and the
test takes a long time to carry out. An easier method is ASTM D-86 distillation. ASTM
D86 distillation is carried out in a glass flask equipped with an electric heater and a water
cooler. The oil sample is heated in the flask and the resultant vapor is condensed and
boiling point (IBP) is defined as the temperature at which the first drop of liquid is
collected. The final boiling point or end point is defined as the maximum temperature
reached during the distillation. ASTM D86 distillation is typically used for light and
distillation (PROU manual, 1996) is similar to ASTM D86 distillation, but is used for
heavier petroleum products and is often carried out under vacuum, sometimes at absolute
One can convert these curves to TBP using correlation (Nelson, 1958, API
handbook, 1970).
ratios o f crude oils as a fimction o f temperature. In this test, the crude oil is continuously
heated without separating the vapor from the remaining liquid (Nelson, 1958). The
temperature versus liquid amount is recorded. To obtain the EFV curve, a series o f runs
at different temperatures are carried out, and each nm constitutes one point (of
10
1.2.2 Gravity, API
API is a density scale used in the petroleum industry. It can be calculated from
In the above equation, specific gravity refers to the weight per unit volume as
compared to water at 60 °F. Most crude have a gravity in the range o f 20 to 45 API. 20
The salts in crude oil are primarily chlorides, sulfates and carbonates o f sodium,
calcium and magnesium. The salt may deposit on heat transfer surfaces and cause
fouling. Chloride salts m ay also decompose to form acids and cause corrosion. The salt
content of the crude oil is required to be less than 10 lb /1000 bbl for crude distillation
(Gray, 1994).
Desalting is performed by mixing the crude oil with 3 to 10 volume per cent water
at temperatures from 90 to 150 °C (200-300 °F). Both the water/oil ratio and the
temperature of operation are functions of the density o f the crude oil. Typical operating
Typical fraction cut points and boiling ranges for atmospheric and vacuum tower
11
Table 1.1 Typical Operating Condition for Desalting
Boiling ranges
Fraction ASTM (°F) TBP (°F) Procesing use
Naphtha ( heavy straight- 180-400 190-380 Reforming stock
run gasoline)
Kerosene 330-540 380-520 Kerosene, jet-50 cut
The main products from a typical crude distillation unit are listed below in the
12
Fuel gas. TT.e fuel gas consists o f methane and ethane. The fuel gas is mainly
used as fuel for refinery furnaces. Sometimes it is also used as feedstocks for ammonia
production.
Wet gas. The wet gas consists o f propane and butane as well as methane and
ethane. The mixture o f propane and butane can be sold as LPG and butanes are raw
catalytic reformate. Naphtha is also used as petrochemical feedstock, mainly for thermal
Gas oils. The light, atmospheric, and vacuum gas oils are used as feedstock for a
hydrocracker or catalytic cracker to produce gasoline, jet, and diesel fuels. The heavy
vacuum gas oils can also be used as feedstock for lubricating oil units.
unit to produce heavy fuel oil or cracking and/or lube base stocks. For asphalt crude oils,
The flash point o f a product is the temperature at which vapor given off will ignite
when an external flame is applied under specified test conditions. A flash point is defined
13
The term gap refers to the difference between the 5% ASTM D86 distillation
temperature o f a heavier product and the 95% ASTM D86 distillation temperature of an
adjacent lighter product. When the distillation curves o f the two products overlap, a
by Nelson (1936). This book discusses the crude distillation in detail. It contains many
charts and experimental data. The 4* edition was published in 1958. In the design
procedure presented in this book, the amount of products is estimated from the crude
distillation curve, then the number o f trays between side withdraws are picked. Next, the
temperature and pressure in each tray are calculated using empirical charts. Heat balance
is then carried out and the reflux is determined. The design procedure ends with the
Another early publication is a paper by Packie (1941). In his method, 5-95 gaps
and 50% distillation temperature differences are used as separation criteria. Packie used
empirical charts to express the relation among the 5-95 gap, the reflux ratio, and the
number o f trays in the section under consideration. However, the empirical nature of
these charts results in inaccuracies in the amount of heat removal and tray temperatures.
Furthermore, as in the case o f Nelson (1936), Packie considered column design and heat
integration separately, that is, heat integration does not start until the column design is
finished.
14
Much later, another book on crude distillation was published by Watkins (1979).
This procedure provided in this book is similar to the one proposed by Nelson (1958). It
uses empirical charts and iterations are required to compute heat and material balances.
However, departing from Nelson and Packie’s criterion regarding heat integration,
Watkins pointed out that “optimizing the crude preheat-tower cooling heat-exchange train
is the heart o f crude unit design, and each case must be studied on an individual basis in
order to arrive at the most economical processing scheme.” The statement reflects the
times the book was written (1979) right after the first energy crisis (1973). However, he
He found that pump-around circuits are only effective to remove heat at higher
temperature. Main steam injection was found beneficial from the point o f view of energy
integration. The combined usage o f large steam injection with smaller overflash was
shown to have large energy savings. In another paper (1998b), he presented a technique
to calculate energy retrofit horizons in process plants. The methodology takes advantage
of two facts: (a) Pinch-type calculations can be performed using operator type
representations and (b) Processes like crude fractionation offer large flexibility in the
The design procedure starts with a sequence of simple columns that are generated by
decomposing the crude main tower. The total number o f trays is assumed to be the same
as that o f Watkins’ design, and the number o f trays for each column is calculated with the
15
assumption that the R/Rmin values are approximately the same for all columns. There is
no thermal coupling between these initial columns. Next, reboilers and thermal coupling
modifications, a tool fi'om Pinch Technology (developed in the eighties), namely, the
grand composite curve, is used. After all the possible design modifications have been
The major advantage o f the procedure presented by Liebmann et al. (1998) is that
it couples the column performance with heat recovery goals. However, the procedure is
not able to assess the trade o ff between steam injection and distribution of heat removal
alternative we present in this paper, is more convenient and straightforward. It does not
rely on any special rules o f thumbs for reflux ratios, and it helps determine better the
Sharma et al. (1999) proposed a method for calculating the maximum pump-
around heat removal. First, a practical minimum reflux ratio for each column section is
determined using Packie’s empirical diagram. Then, the heat removal in the upper part o f
the column is calculated using a heat balance. The upper part may start fi'om an arbitrary
tray and end with the condenser. Following, the upper part is extended tray by tray and
heat siuplus is calculated for each tray. The resultant heat surplus data are used to
construct a column grand composite curve. Finally, the maximum heat removal for each
section is determined using the column grand composite curve. A major advantage o f this
method is that the maximum heat removal can be estimated quickly without the need o f
16
simulation. However, because Packie’s diagram is empirical and the effect o f the
stripping steam is not included, the heat removal calculated is not accurate.
simulator and heat integration procedures as means o f developing a new design procedure
that would capture rigorously the column behavior, the effect o f steam and refluxes and
the relationship with heat integration. Bagajewicz (1998b) suggested the use of rigorous
This dissertation discusses a rigorous procedure that uses simulators that are able
parameters and complements it with state of the art heat exchanger network design
presented. This procedure uses a heat demand-supply diagram and commercial software
to determine the energy target for a plant processing several crude oils. This procedure
allows each crude to be processed at optimal conditions. Chapter 3 addresses the problem
o f designing a crude distillation with a pre-flash imit. A rigorous definition for “carrier
effect” is provided and the role o f light components in the crude distillation is analyzed in
detail. Then a rigorous procedure for energy targeting in the design o f crude distillation
17
distillation which was claimed by Liebmann and Dhole (1995) to be more energy-
efficient than the conventional design. The impact o f the different heating patterns in the
stripping-type design is therefore analyzed and compared with the conventional design.
Chapter 5 addresses the energy targeting for the design o f complete distillation plants,
that is, including both the atmospheric tower and the vacuum tower. Chapter 6 focuses on
the design of a multi-period heat exchanger network for a complete distillation plant. The
plant has a preflash drum to facilitate the processing o f light crudes. The last chapter
(Chapter 7) discusses the possible alternative designs and presents preliminary results.
1.6 References
1. American Petroleum Institute, 1970, Technical Data Book - Petroleum Refining, 2nd
2. American Petroleum Institute, 1970, Technical Data Book - Petroleum Refining, 2nd
4. Bagajewicz, M. J., Energy savings horizons for the retrofit o f chemical processes.
(1), 1998b
5. Gary, James H., Petroleum refining : technology and economics, 3rd ed. M. Dekker,
1984
18
7. Liebmann, K., Integrated crude oil distillation design. Ph. D. dissertation. University
8. Miller, W., Osborne, H., The science o f petroleum. Volume II, Oxford University
York, 1936.
11. Packie, J.W. Distillation Equipment in the Oil Refining Industry. AIChE
13. Rhodes, A. K., Worldwide refining report. Oil & Gas Journal, December, 37-86,
1993.
1979.
19
Chapter 2 Rigorous Targeting Procedure for the Design of
2.1 Overview
design. It has the added complexity that these units should be able to process different
types o f crude, sometimes from heavy to light. Important heat exchange also takes place,
and the energy efficiency is related to the column design parameters. Part I o f this two-
part paper presents a rigorous targeting methodology to design this multipurpose plant,
which can be implemented using a commercial simulator. Part II deals with the heat
2.2 Introduction
the crude processed. The conventional design (Figure 2.1), consisting o f a column with
side strippers and pump-around circuits, appeared 70 years ago (Miller, 1938), and is still
the design used in the refining industry. Watkins (1979) proposed a design procedure for
this system and discussed a few variants such as pump-back reflux and stripping using
reboilers. A few alternative designs can be found in the literature. For example, the
Brugma (1941) and is being used in several industrial sites. Another old design, the
carrier design, was proposed as early as the 1920s. This design makes use o f light
components to enhance the separation in the stripping section o f the column. Nelson
20
(1958) also mentioned some other alternative designs. All these old designs have been
abandoned for reasons that are not completely known or understood. One important fact
is that they were abandoned before the seventies, when energy consumption started to
play an important role in process economics. Because energy efficiency is now desired,
all these designs merit réévaluations. Nevertheless, the conventional design is widespread
and popular.
CONDENSER
w ater
• n a p h th a
PAl
PA2 kerosene
ste am
PA3 diesel
crude DESALTER
h e n (~
gas oil
sour w ater; steam
HEN FURNACE re sid u e
Crude is mixed with water and heated in a heat exchanger network before
entering a desalter where most o f the water containing the salt is removed. The desalted
crude enters another heat exchanger network and receives heat from hot streams. Both
heat exchanger networks make use o f the vapors o f the main column condenser, the
pump-around circuit streams, and the products that need to be cooled. The preheated
crude then enters the furnace, where it is heated to about 340-370 °C. The partially
21
vaporized crude is fed into the flash zone o f the atmospheric column, where the vapor
and liquid separate. The vapor includes all the components that comprise the products,
while the liquid is the residue with a small amount o f relatively light components in the
range o f gas oil. Tliese components are removed from the residue by steam stripping at
the bottom o f the column. In addition to the overhead condenser, there are several pump-
around circuits along the column, where liquid streams are withdrawn, cooled, and sent
back to upper trays. Products are withdrawn in liquid state from different trays and then
offered a detailed discussion o f the effect o f the different variables on the energy
Crude oil is a complex mixture. There exist about one thousand distinguishable
components with boiling temperatures varying from room temperature to over 550 °C.
Crude distillation yields mixtures called naphtha, kerosene, diesel, and gas oil. These
• Large processing quantity: The charge rate is the largest among all petroleum or
chemical processing units. The typical processing capability is around 15,000 per
day (100,000 bbl/day). In such a large-scale process, energy cost accounts for a larger
between the top tray and the flash zone is about 250°C, which means large heat
22
• Absence o f a reboiler: The main column functions as a rectifying section for products,
rather than component fractions as in the discrete component separation case. The
because all components constituting the light product have to travel through trays
The major objective in the design of crude distillation units is to find the most
energy efficient separation structure. Although some ideas exist for the design o f energy
integrated distillation schemes (Agrawal, 1996), they are not directly applicable to crude
• The number of components in the crude is too large to handle. Usually around 30 to
• Previous separation sequencing studies assumed the products are pure, however
Packie (1941) pioneered the field o f crude fractionation design. In his method, 5-
95 gaps and 50% distillation temperature differences are used as separation criteria. The
term gap refers to the difference between the 5% ASTM D86 distillation temperature of a
heavier product and the 95% ASTM D86 distillation temperature o f an adjacent lighter
product. When the distillation curves o f the two products overlap, a negative gap appears.
Packie used empirical charts to express the relation among the 5-95 gap, the reflux ratio,
23
and the number o f trays in the section under consideration. However, the empirical nature
o f these charts results in inaccuracy and prevents optimal designs. Furthermore, in his
design procedure, Packie considered column design and heat integration separately. Heat
Watkins (1979) pointed out that “optimizing the crude preheat-tower cooling
heat-exchange train is the heart o f crude unit design, and each case must be studied on an
individual basis in order to arrive at the most economical processing scheme.” However,
The design procedure starts with a sequence o f simple columns that are generated by
decomposing the crude main tower. The total number o f trays is assumed to be the same
as that o f Watkins’ design, and the number o f trays for each column is calculated with the
assumption that the R/Rmin values are approximately the same for all columns. There is
no thermal coupling between these initial columns. Next, reboilers and thermal coupling
are introduced in order to reduce utility consumption. The grand composite curve is used
to assess the proposed modifications. After all the possible design modifications have
been explored, these columns are merged into a single complex column.
The major advantage o f the procedure presented by Liebmann et al. (1998) is that
it couples the column performance with heat recovery goals. However, the procedure is
not able to assess the trade o ff between steam injection and distribution of heat removal
alternative we present in this paper, is more convenient and straightforward. It does not
24
rely on any special rules o f thumbs for reflux ratios, and it helps determine better the
Sharma et al. (1999) proposed a method for calculating the maximum pump-
around heat removal. First, a practical minimum reflux ratio for each column section is
determined using Packie’s empirical diagram. Then, the heat removal in the upper part o f
the column is calculated using a heat balance. The upper part may start from an arbitrary
tray and end with the condenser. Following, the upper part is extended tray by tray and
heat surplus is calculated for each tray. The resultant heat surplus data are used to
construct a column grand composite curve. Finally, the maximum heat removal for each
section is determined using the column grand composite curve. A major advantage o f this
method is that the maximum heat removal can be estimated quickly without the need o f
simulation. However, as Packie’s diagram is empirical and the effect o f the stripping
steam is not included, the heat removal calculated is not accurate. In contrast, the
procedure presented in this paper is based on rigorous simulations and can capture the
relationships between the column variables and the heat integration opportunities.
In this paper we present a new procedure for crude distillation design. The major
Heat demand-supply diagrams are used as a tool guiding the design. The major
advantage o f the heat demand-supply diagram is that the contribution o f each process
stream or pump-around to the total utility consumption is shown explicitly. This feature
25
• The interaction between steam stripping and pump-around duties is taken into
consideration.
• The starting point o f the column design is a complex column without pump-around
circuits.
While general procedures that would render globally optimal solutions are a
desirable goal, there is also interest in determining the optimal parameters for the subset
o f conventional units. This choice is made because practitioners are not always willing to
make radical departures from this design. In addition, the knowledge o f such optimal
designs provides a useful horizon for retrofit procedures. In addition, the heat exchanger
network design procedure presented in part II renders a structure suitable for the
tool, is presented first and the roles o f the different column design variables in this
diagram are discussed. Next, limitations in the pump-around circuit heat load are
described. Finally, the design targeting procedure is presented. This rigorous procedure is
based on the use o f commercial simulators, departing from the use of charts, rules o f
enthalpy diagrams (Hohmann, 1971; Huang and Elshout, 1976; Naka et. al., 1980;
Andrecovich and Westerberg, 1985; Terranova and Westerberg, 1989; Dhole and
26
LinnhofF, 1993). In the demand-supply diagram, a stream is represented by a curve. This
curve represents the product of mass flowrate and specific heat capacity (true or apparent
demand-supply diagram for typical crude fractionation units, like the one o f Figure 2.1, is
In setting up the diagram, a heat demand line is first drawn and used as a
background. The crude is the only cold stream. In some cases, as proposed by Liebmann
et al. (1998), water at room temperature to produce steam is also considered a cold
considered as a cheap or even free heat source. To locate the hot streams (heat supply),
the usual minimum temperature difference is used. Thus the temperatures o f hot streams
are shifted to the left by this minimum difference, which has been traditionally named
heat recovery minimum approximation temperature (HRAT). The area below the heat
demand line represents the total heat demand o f the unit without heat recovery.
When the supply exceeds the demand, one can move the surplus part o f the
supply to a lower temperature region where the supply is deficient. Figure 2.2 shows two
areas where the supply is in deficit (gray areas). The left gray area can be covered by the
heat surplus fi'om the condenser or from PA l. The right one can be covered using the
excess o f PA2. This illustration is omitted throughout the paper, assuming that this area
matching is implicit.
27
A
.2
.0
o
o
PINCH
0.8 PA1
CRUDE
0.6 COND
Q.
o
*
0.4
PA2 PA 3
0.2
RES
0
100 200 300 400
PRODUCTS
TEMPERATURE, °C
1. Naphtha and condensed water. 2. Sour water from the desalter. 3. Pump-around 1. COND: condenser.
PA: pump-around. RES: residue. Products (from top to bottom): kerosene, diesel and gas oil.
The location of the pinch point can be easily obtained from this diagram. It is the
lowest temperature at which the demand is larger than the supply after the shifting and
area matching has been performed. Finally, the heating utility is given by the unmatched
demand on the right, and the cooling utility is given by the extra supply on the left.
supply and improving the match between the supply and the demand.
28
• Decrease o f demand
a) A decrease in heat demand can be realized by moving the demand line down, that is,
decreasing the flowrate. One way o f doing this is to flash the crude at lower
temperatures and send the vapor to a tray above the flash zone. In practice,
vaporization before the furnace inlet is pressure suppressed to avoid two-phase flow.
b) Another way to decrease heat demand is to reduce the target temperature of the crude.
This can be achieved by lowering the pressure drop from the outlet o f the furnace to
the overhead reflux drum o f the column. In this sense, a vacuum operation is even
better, but it is excluded for other reasons (mainly cost). Another way of decreasing
the final temperature is using larger amount o f steam, but the introduction o f steam
paper.
Withdrawing certain products in the vapor phase instead o f in a liquid phase has
the advantage that condensation heat is released at a higher temperature. This option is
not explored in this paper, mainly because it is a major departure from the conventional
scheme.
29
Assume there is a large heat surplus in a moderate temperature range and a heat
deficit in a higher temperature range. One way to improve this mismatch is to move a
part o f the heat surplus to a higher temperature by increasing the duties o f the pump-
around circuits. The design procedure proposed in this paper relies on the idea of
distributing heat among the condenser and pump-around circuits. We focus on this
procedure next.
The original purpose o f adding pump-around circuits was to reduce vapor and
liquid traffic at the top section o f the column (Watkins, 1979). Without pump-around
circuits, all condensation heat has to be removed from the condenser, which results in a
large vapor flowrate at the top trays. We explore now the limit o f heat that could be
the maximum pump-around duty, we carry out a heat balance for this envelope.
= '%+Zie/II
+C ■A" + ■ K ,
In equation (1.1), V ^, are the steam flowrate at the flash zone and the
hydrocarbon vapor flowrate at the flash zone respectively, v f , y ° are the steam
flowrate at tray j and the hydrocarbon vapor flowrate at tray j respectively. In Figure 2.3,
30
we use Vjn = and Vj = V j +V f . It is assumed that water is insoluble in
liquid streams.
(1 2 )
ie.1
y" + iE
e // /
f.,
(1.3)
1'° = i . - . + Z f .
le//
There are six terms on the right hand side o f equation (1.4). From left to right,
these terms represent condensation heat o f the overflash stream Lq, condensation heat of
the products leaving envelope III, apparent heat released by the hydrocarbon vapor V ° ,
and apparent heat released by the steam streams. The sixth term stands for tlie
vaporization heat o f internal reflux Lj.i. Apparently, when Lj.i goes to zero, the heat
removal from envelope III reaches its maximum. By including more pump-arounds in
envelope III and applying equation (1.4) accordingly, one can find the maximum heat
It can be shown through an overall heat balance that the total amount o f heat to be
removed fi-om the column depends on the yields o f the products. In addition, shifting heat
31
from envelope I I to envelope III results in a decrease of Lj.i. Thus, the shifting can take
^ Water
PAl Naphtha
SK
Kerosene
PA2
" III
SD
PA3 "► Diesel
SG
^ Gas oil
Hot Crude FZ
SR
Residue
It is well known that heat shifting reduces separation efficiency. The presence o f
the pump-around circuit decreases the number of effective ideal trays (Bagajewicz,
1998). The effect can be even more detrimental to separation if the flowrate of a pump-
around circuit is increased. As we shall see later, these effects can be compensated to a
32
certain extent by increasing the steam rate in the side strippers. Another solution is to
increase the number o f trays. However, with the total number o f trays kept constant, the
atmospheric column. First, the Watkins design method is used to obtain an initial scheme
the direction o f heat shifting needed for maximum energy efficiency is determined. This
procedure is repeated for at least the lightest crude and the heaviest cmde that will be
processed. Thus, the design procedure is divided into two parts, the targeting procedure
and the multipurpose heat exchanger network design. This paper focuses on the targeting
procedure, which is presented next. After this, the goals o f the heat exchanger network
design procedure are outlined. The heat exchanger network design procedure is presented
in part II.
Step 1: Begin with the lightest crude to be processed. As the lightest crude has the
highest yields o f light distillates, the supply o f heat is the largest. Next, the major design
parameters (the number o f trays in each section, the pressure drop, and the amount o f
stripping steam) are chosen using the guidelines offered by Watkins with one exception:
33
Step 2: The simulation is performed next. Usually the column is not difficult to converge,
the region between the top tray and the first product withdrawal tray. The location o f the
pump-around circuit withdrawal and the return temperature are conveniently chosen so
that the energy recovery is maximized. This is discussed further when presenting the
example.
Step 5: If the product gap becomes smaller than required, the stripping steam flowrate is
to be increased to fix the gap. As long as the steam added has a lower cost than the
energy saved, one can continue shifting loads. Otherwise, it is advisable to stop when a
Step 6: If there is heat surplus from the pump-around circuit just added, transfer the heat
to the next pump-around circuit between draws in the same way as in step 4. If not, stop.
At this stage, once this procedure is repeated for different crudes, one is left with
heat removal targets fi’om the condenser, the products and several pump-around circuit
streams. Typically, since the light crude is the one that needs a larger reflux, it exhibits a
larger amount o f pump-around circuit duties. After these targets are determined, it is
shown that there is still some flexibility to move heat fi’om one pump-around to another, a
feature that may be helpful in the final design o f the heat exchanger network, or for
retrofit. The above procedure is illustrated first. The results o f this targeting procedure are
34
used as motivating material to discuss the goals o f the multipurpose heat exchanger
2.6 Illustration
The properties o f the light crude, intermediate crude and heavy crude are shown
in Tables 2.1, 2 and 3. Table 2.4 indicates the specifications of the products. The product
withdraw locations are detemiined according to W atkins’ guidelines and the results are
35
Tabic 2.3 Light-ends Composition o f Crude in volume per cent
Compound Light Crude Intermediate Crude Heavy Crude
Ethane 0.13 0.1 0
Propane 0.78 0.3 0.04
Isobutane 0.49 0.2 0.04
n-Butane 1.36 0.7 0.11
Isopentane 1.05 0 0.14
n-Pentane 1.30 0 0.16
Total 5.11 1.3 0.48
There are 34 trays in the main column and 4 trays in each stripper. The flowrates
of stripping steam streams are estimated and adjusted to 10 lb per barrel o f product, as
36
suggested by Watkins. The total energy consumption (£) is calculated using the
following expression:
E = U^0.1 * (2.5)
where U is the minimum heating utility obtained using straight pinch analysis, and
^ /// is the summation o f energy flow of all steam streams. Because low-pressure
steam is cheaper than fuel gas on the same amount o f heat content, a weight factor of 0.7
is used for the steam. The total energy consumption is used as an objective function.
Simulation results for the initial scheme with no pump-around circuits are shown
in Table 2.6. Note the product gaps are well above the specifications.
shown in Figure 2.4. There is huge heat surplus in the condenser region, which results in
a large cooling utility. Meanwhile, a large heat deficit exists above 155 °C. As the total
heat supply is almost constant, the way toward energy savings is to change the heat
supply profile. That is, instead o f supplying all heat at a low temperature, some heat can
37
be supplied at a higher temperature where the heat demand is larger than the heat supply.
In other words, transfer some heat from the condenser to a pump-around circuit as
1.4
PINCH
O
COND
0.8 -
CRUDE
Q 0.6
^ 0.4 -
RES
Figure 2.4 Heat Demand-Supply Diagram for Crude Distillation without Pump-Around
Circuits
If a pump-around is above all side-withdrawal product lines, the heat that can be
transferred from the condenser will be the maximum. Therefore, the first pump-around
38
has to be above the kerosene withdrawal tray. The question is how many trays one should
put between the condenser and the top pump-around region. We recommend the top
based on the observation that the trays below a product withdraw line and above an
adjacent pump-around circuit receive little reflux and barely contribute to separation. The
pump-around stream is withdrawn from tray 4, cooled in the heat exchangers and
returned to tray 2. The return temperature is 104.4 °C, which is optimized after the duty is
determined.
The duty o f the top pump-around (PA l) is increased gradually and product gaps
are examined in each simulation. The kerosene-naphtha gap decreases with the increase
o f PAl duty, but remains well above that o f specification, while the other gaps are almost
unchanged. The heat shift continues without violating the gap specifications until the
reflux ratio is around 0.1. Further heat shift would result in liquid drying up on the top
tray. Thus, the limit o f the heat shifting has been reached. The duty of 62 MW represents
the total amount o f heat one could obtain from all pump-around circuits. The following
steps consist of distributing this amount of heat properly among several pump-around
circuits. The main operation variables of the scheme with one pump-around are shown in
Table 2.6.
around scheme.
• The kerosene-naphtha gap is reduced from 25 °C to 23 °C, remaining well above the
39
Table 2.6 Comparative Results o f One Top Pump-Around and No Pump-Around
Product No Pump-around One Pump-around
Naphtha Flowrate 250 m^/hr 248 m^/hr
Kerosene Flowrate 144 m^/hr 146 m^/hr
Diesel Flowrate 70 m^/hr 70 m^/Tu"
Gas Oil Flowrate 121 m^/hr 121 m^/hr
Residue Flowrate 211 m^/lir 211 m^/hr
Kerosene Stripping Steam Ratio* 9.82 9.68
Diesel Stripping Steam Ratio 10.22 10.27
Gas Oil Stripping Steam Ratio 10.12 10.11
Residue Stripping Steam Ratio 10.19 10.19
Kerosene-Naphtha (5%-95%) 25.12°C 23.0°C
Gap
(5-95) Diesel-Kerosene Gap 5.14°C 5.3 r c
(5-95) Gas Oil- Diesel Gap 0.93 "C 0.91°C
Kerosene Withdrawal Tray 238.8°C 237. r c
Temperature
Diesel Withdrawal Tray 298.7°C 298.7°C
Temperature
Gas Oil Withdrawal Tray 338.7°C 338.7°C
Temperature
Residue Withdrawal Temperature 347.8°C 347.8°C
Condenser Duty 103.86 MW 41.70 MW
Condenser Temperature Range 155-43.3 °C 146.4-43.3 °C
Pump-around 1 Duty - 62.14 MW
Pump-around 1 Temperature - 179.6-104.4 °C
Range
Flash Zone Temperature 358.6 °C 358.6 °C
Energy Consumption (E) 103.78 MW 96.77 MW
*Steam amount in lb/hr over the amount of product in bbl/hr.
40
• The yield o f naphtha decreases and the yield o f kerosene increases. This is because
some light components of the vapor are absorbed by the cold pump-around stream
and carried to the kerosene withdrawal tray. Note that the total yield of the two
We now turn our attention to the resulting heat demand-supply diagram (Figure
2.5). The shaded area is the energy savings achieved by adding PA l. The heat surplus in
because the heat surplus is larger than the demand below the PA l withdrawal
The second pump-around (PA2) is positioned between tray 10 and tray 12, just
would result in heat surplus in the region o f P A l, while a very high return temperature
would not alter the energy savings but result in a heavier liquid traffic in the PA2 region.
With the increase o f the PA2 duty, the gap between kerosene and naphtha decreases
quickly. Table 2.7 shows the change o f gaps as a function of the duty of pump-around
PA2.
41
1.4
PINCH
O
0 .8 - PA1 CRUDE
& 06. -
0.4
COND
0 .2 -
RES
Figure 2.5 Heat Demand-Supply Diagram for Crude Distillation with a Top Pump-around
Two Pump-Aroimd Circuits
When the duty o f PA2 is larger than 33.7 MW, the kerosene-naphtha gap does not
satisfy the specification. To recover this gap, one could increase the stripping steam
flowrate or increase the number of trays in the naphtha-kerosene section. The former
option is used in this work, while the latter may not be sufficient or even practical on its
own. The kerosene and diesel stripping steam fiowrates are adjusted with a controller in
42
Table 2.7 Effect o f Increasing PA2 Duty without Changing Steam Fiowrates
1 2
Duty o f PA2 29.31 MW 33.71 MW
Duty o f PA l 32.83 MW 28.43 MW
Duty o f Condenser 41.94 MW 42.03 MW
(5-95) Kerosene-Naphtha Gap 18.49 °C 16.60 °C
(5-95) Diesel-Kerosene Gap 1.63 °C 1.48 °C
(5-95) Gas Oil- Diesel Gap 1.22 °C 1.23 °C
Energy Consumption 70.59 MW 67.35 MW
With the help o f the stripping steam, it is possible to move more heat from PAl to
PA2. The trade off between increasing energy recovery and spending more steam is
evaluated using equation (2.5). Heat shifting continues until the liquid reflux at the
kerosene withdrawal tray is small and /or the kerosene-naphtha gap cannot be recovered
even with increased amounts o f stripping steam. This is a limit imposed by the separation
requirement. The limiting case is shown in Table 2.9 (first column) and should be
The major changes from one pump-around to two pump-around circuits are:
• The flowrate o f the kerosene stripping steam is nearly doubled. The large extra steam
stream. The top section o f the column becomes less hot because o f the increased
43
The heat demand-supply diagram (Figure 2.6) shows a good match, and the pinch
temperature increases to the value o f the PA2 withdrawal temperature. The heat surplus
in the region o f PA l is still high, but further shifting would cost too much steam to be
beneficial. Therefore, this remaining heat surplus is useless. Now the only heat surplus
transferable is located in the PA2 circuit, shown as the shaded area in Figure 2.6. To
make use o f this heat surplus, it is necessary to add a third pump-around circuit.
The third pump-around (PA3) is located between tray 17 and tray 19. The return
temperature is 232 °C. Heat is shifted gradually from PA2 to PA3, with the gaps
maintained by adjusting steam fiowrates. The effect o f the duty of PA3 on energy
consumption is shown in Table 2.8. A summary o f all variables is given in Table 2.9.
44
1.4
1.2
1 PINCH
o
g 0.8
CRUDE
5
cL 0.6
O COND
2 0.4
0.2
0
0 100 200 300 400
TEMPERATURE, °C
Figure 2.6 Heat Demand-Supply Diagram for Crude Distillation with Two Pump-Around
Circuits.
At the beginning, the energy consumption decreases with the increase o f the duty
o f PA3. However, when the PA3 duty exceeds 8.8 MW, the energy consumption stays
constant in a rather wide range (Table 2.8). This is because little heat surplus exists in the
region of PA2. Therefore, more heat shift makes no difference. Beyond this stable range,
more heat shift to PA3 results in an increase in energy consumption due to increased use
o f steam, which means that the cost of additional steam consumption outweighs the gain
in energy recovery. Clearly 8.8 MW is the right point to stop. This effect cannot be
45
Table 2.9 Comparative Results for Two and Three Pump-Around Circuits
46
1.4
PINCH
CRUDE
COND
0.4
PA2 PA3
0.2
RES
TEMPERATURE, °C
Figure 2.7 Heat Demand-Supply Diagram for Crude Distillation with Three
Pump-Around Circuits
Figure 2.7 is the heat demand-supply diagram. The heat surplus previously in the
region o f PA2 (Figure 2.6) has been moved to the PA3, which accounts for the decrease
in energy consumption.
remarkable effect on the column temperature profile. Figure 2.8 shows stream
temperature changes as a function o f the duty o f the third pump-around. With the
47
increase o f these duties, the temperatures o f the products and the pump-around circuits
decrease.
350
Ü 300
o
UJ
a:
250
0:
LU
CL
UJ
h- 200
150
100
0 5 10 15 20 25 30
D U T Y . MW/ °C
the pinch temperature is located in the region o f PA2, its return temperature affects the
48
1.2
1
PAl
0.8
CRUDE
0.6
O
o COND
0.4
Q. RAI
U
♦
0.2 PA2
RES
0
TEMPERATURE, °C
In the figure, PA3 is not shown for simplicity. The dotted line is the new PA2
region with a higher return temperature and the shaded areas are the changes incurred.
When the PA2 duty is constant, a higher return temperature results in a lower withdrawal
temperature. The total effect depends on the trade off between these two effects, or the
area difference between the shaded triangle and the shaded rectangle. The trade off is
shown in Table 2.10. The optimal return temperature is found to be 177.8 °C.
49
Table 2.10 Effect o f the Return Temperature o f PA2 on Energy Consumption
At this point we have reached the best scheme for the light crude. Next we
Heavy Crude
The total energy consumption and the pump-around duty distribution are shown
in Table 2.11. The heat demand-supply diagram and the operation variables for a scheme
with three pump-around circuits are shown in Figure 2.10 and Table 2.12. The following
50
• The energy consumption changes very little when shifting heat from the condenser to
the pump-around circuits, especially when heat is shifted from PAl to PA2 or PA3.
• This is because that there is no heat surplus in the condenser region (Figure 2.10).
However, because the light crude and the medium crude require the PA2 and PA3
heat exchangers, shifting heat from PAl to PA2 and PA3 in heavy crude design may
be necessary.
0.8
0.6 CRUDE
PA3
Ü
o
0.4
RES
O
Q. COND
♦ 0.2
PA1
SW PA2
0
100 200 300 400
TEMPERATURE, °C
51
Table 2.12 Results for Heavy Crude
Product Heavy Crude
Naphtha Flowrate 55.37 m^/hr
Kerosene Flowrate 48.64 m^/hr
Diesel Flowrate 69.36 m^/hr
Gas Oil Flowrate 29.37 m^/hr
Residue Flowrate 592.51 m^/hr
Kerosene Stripping Steam Ratio* 1.63
Diesel Stripping Steam Ratio 2.98
Gas Oil Stripping Steam Ratio 37.9
Residue Stripping Steam Ratio 2.68
(5-95) Kerosene-Naphtha Gap 26.07 °C
(5-95) Diesel-Kerosene Gap 0.86 °C
(5-95) Gas Oil- Diesel Gap -5.84°C
Kerosene Withdrawal Tray Temperature 259.7°C
Diesel Withdrawal Tray Temperature 317.4 °C
Gas Oil Withdrawal Tray Temperature 344.4 °C
Residue Withdrawal Temperature 366.7 °C
Condenser Duty 14.8 MW
Condenser Temperature Range 123.3-18.5 °C
PAl Duty 20.8 MW
PAl Temperature Range 175.7-104.4 °C
Flash Zone Temperature 353.2 °C
Energy Consumption 81.49 MW
When heat is shifted to PA2 and PA3, more steam is needed for the diesel stripper to
regain the kerosene-diesel gap. The diesel stripping steam fiowrates for the designs with
one pump-around, two pump-around and three pump-around circuits are 32.5, 48.5 and
52
113.4 kg-moie/hr respectively. Although the steam consumption increases, the total
energy consumption is barely affected because the heat from the extra steam is utilized to
• The separation o f kerosene and diesel in the column is much easier than that o f the
light crude. Before stripping, the gap between kerosene and naphtha is 17.2 °C,
53
Note that for the light crude, PA3 duty increases with the increase o f HRAT. This
can be explained using the heat demand-supply diagram (Figure 2.7). When the HRAT is
5.6°C, there is almost no heat surplus in the region o f PA2. However, when HRAT is
increased, the crude demand curve is moved to the right and heat surplus appears again.
Thus, the heat surplus needs to be reduced to achieve the maximum energy savings. The
heavy crude behaves differently. As there is no heat surplus in the region of the
condenser and P A l, shifting heat fi'om PA l to PA2 or PA3 does not reduce the net heat
demand while more stripping steam is needed to keep the product gaps. At low HRAT
(e.g., 5.6 °C), most o f the heat coming from the condenser can be used because o f the
heat deficit in the condenser region. However, when HRAT is raised, the overlapping
between the crude curve and the condenser curve reduces, and part o f the heat from the
condenser is at a temperature that is too low to be usable. In such a case, the heat from
the increased steam cannot be used. Therefore, heat shifting to the lower pump-around
These calculations were also performed for the intermediate crude (Table 2.15).
In this case, the heat distribution does not change with HRAT This is because there is
always a heat surplus in the region of PA l and a heat deficit in the region o f PA2. The
54
heat surplus in the region o f PA l prompts maximum heat shift to PA2, while the heat
deficit in PA2 excludes the need for shifting heat to PA3. Thus, the optimal solution is to
2.8 Flexibility
The basis o f this design procedure is the transferring o f heat duty from the
condenser to the lower pump-around circuits. In doing so, limits to this transfer are
encountered. In addition, heat surplus can be transferred back. For example, the heat
surplus observed in the pump-around PA l for the light crude (Figure 2.7) could in
principle be transferred back to the condenser without affecting the utility consumption.
2.9 Reboiler
Reboilers are rarely used in the conventional crude distillation because their
installation is expensive and sometimes they are less efficient than steam stripping.
However, Liebmann et al. (1998) suggested that the use o f reboilers can lead to energy
savings. This conclusion is made by analyzing a scenario where water to produce steam
If one assumes that steam is another heat sink, then the location o f the pinch
changes. Figure 2.11 shows the demand-supply diagram of the light crude. The heavy
solid curve represents the heat demand for both the crude and the steam. The heat
demand for the crude is shown as a broken curve for comparison. In this figure, a heat
55
1.4 -1
1.2 - PINCH
I
o 1 -
o
0.8 -
i DEMAND
CL 0.6 - COND
o 1
"K 258°C
0.4 H
182 °C
0.2
0 -
300
200 400
0 K E ^O
TEMPERATURE, C
igure 2.11 Heat Demand-Supply Diagram for Light Crude Including Steam as Cold
Stream
Each option has its advantage. Shifting the heat surplus to the third pump-around
can reduce the duty o f the furnace. Heat provided by the furnace is relatively expensive,
because the furnace is usually less efficient than a boiler. However, heat shifting from
PA2 to PA3 deteriorates the separation and consequently, extra steam is needed for
56
stripping gas oil. Thus, the choice between the first two options is a matter o f a cost-
beneflt analysis, and it can not be determined merely on energy savings considerations.
The choice between producing steam versus installing a reboiler depends on the
amount o f light components to be stripped. The use o f steam is more efficient for
stripping a low amount o f light components, while a reboiler is better for removing a
relatively large amoimt o f components (Nelson, 1958, Liebmann et ai, 1998). It is also
possible that a combination o f these options offers an optimal solution. To find the best
allowing its production to participate in the design procedure, then the installation of
reboilers should be analyzed using the demand-supply diagram offered in Figure 2.6 or 7.
In this case, the shifting o f heat to PA3 is energy savings related and the use of reboilers
needs to be made by using some other external heat source because process heat is no
longer available in this case. Thus, their installation can only influence the steam
consumption but it is replaced by a similar duty. The choice is therefore not driven by
energy savings, but by a cost-benefit analysis that depends on the cost o f the reboilers to
The light crude and the heavy crude represent two extremes in the raw material
spectrum. Any other crude could be thought o f as a mixture o f the two crudes. Based on
the above information, we may design an optimal HEN that allows several crudes being
processed at optimal conditions. The conjecture is that building a network addressing the
57
extremes allows also processing a crude o f intermediate density at maximum energy
2.11 Conclusion
In this paper, a rigorous targeting design procedure has been proposed for the
the existing procedures for several reasons. First, this procedure aims at finding the best
scheme for a multipurpose crude distillation unit that processes a variety o f crudes.
grand composite curves are used as a guide directing the search for optimal schemes. An
advantage o f heat demand-supply diagrams is that the role of each stream, heater or
cooler in the total energy consumption is clearly shown, so the search of the best scheme
instead o f simple assumptions. The second part paper concentrates on the design o f heat
exchange network.
2.12 Nomenclature
58
/ / / = enthalpy o f stripping steam i, MW
R = reflux ratio
RES = Residue
S = steam
59
= water (steam) flowrate at flash zone
2.13 References
2. Andrecovich, M., and Westerberg, A., A Simple Synthesis Method Based on Utility
(1985).
5. Brugma, A. J., The Brugma Process. Refiner and Natural Gasoline Manufacturer,
20(9), 86,(1941).
6. Dhole, V. R. and Linnhoff, B., Distillation Column targets. Computers and Chemical
60
7. Glinos, FC., and Malone, M. F., Optimality Regions fo r Complex Column Alternatives
9. Huang, P., and Elshout, R., Optimizing the Heat Recovery o f Crude Units. Chemical
10. Ji S. and M. Bagajewicz. Rigorous Targeting Procedure For The Design O f Crude
11. Liebmann, K., and Dhole, V. R., Integrated Crude Distillation Design. Computers &
12. Liebmann, K.; Dhole, V. R. and Jobson, M., Integration Design O f A Conventional
13. Linnhoff, B., Dim ford, H., and Smith R., Heat Integration o f Distillation Columns
14. Miller, W. and Osborne, H. G. History and Development o f Some Important Phases
15. Nelson, W. L., Petroleum Refinery Engineering, 4 ^ ed.. New York, (1958).
16. Packie, J.W., Distillation Equipment in the Oil Refining Industry. AIChE
61
17. Naka, Y., Terashita, M., Hayashiguchi S., and Takamatsu T., An Intermediate
18. Sharma, R., Jindal, A., Mandawala, D., and Jana, S. K., Design/Retrofit targets o f
1374-1379, (1989)
20. Tedder, D.W., The Heuristic Synthesis and Topology o f Optimal Distillation
21. Watkins, R. N., Petroleum Refinery Distillation. G ulf Publishing Company, (1979).
62
Chapter 3 Rigorous Targeting Procedure for the Design of
3.1 Overview
This paper presents a systematic procedure to obtain design targets for heat
pre-flash drums. It is shown that under the same high product yield conditions, pre-
fractionation or pre-flashing are not advantageous from the energy point of view. This is
in great part due to the loss o f the carrier effect that light components have in separating
heavy gas-oil fractions in the flash zone. However, if one accepts the yield of
less energy.
3.2 Introduction
In previous work (Bagajewicz and Ji, 2001), a systematic procedure for the design
starts with a column without pump around circuits and as heat is transferred from the
condenser to pump around circuits with higher temperature, a trade off between steam
usage and furnace savings is established. This transfer o f heat is possible due to the well-
known operating and design flexibility that crude fractionation installations exhibit,
knowledge that was formalized in detail by Bagajewicz (1998). The procedure presented
by Bagajewicz and Ji (2001) makes use o f rigorous simulations and heat supply-demand
63
diagrams similar to those introduced by Andrecovich and Westerberg (1985) and
Terranova and Westerberg (1989). Based on these targets, a heat exchanger network
developed (Bagajewicz and Ji, 2001; Bagajewicz and Soto, 2001). These design
3 __ condenser
PAl
heavy naphtha
light naphtha
steam
Crude oil
desalter steam
diesel
M
sour water
furnace ► residue
prefractionation
tower mam tower
In the pre-flash scheme (Figure 3.2) the intended effect is to avoid unnecessary
appropriate tray in the column for further fractionation. In the pre-fractionation scheme
64
(Figure 3.1), light products obtained in the pre-fractionation column are not sent to the
main column.
condenser
naphtha
PAl
steam
PA2
Crude Oil
vapor
c kerosene
desalter steam
PA3 ^
diesel
steam
sour water gas oil
preflash drum steam
furnace residue
There are several unresolved questions regarding these designs. Among others:
1. In the case of pre-flashing, what is the optimum temperature o f the flash drum?
3. In the case o f pre-flashing, what is the feed tray that one needs to use to feed the
4. What are the loads o f the pump-aroimd circuits and what are the steam flowrates for
In addition, both schemes avoid introducing lights in the feed tray (flash zone)
with the aforementioned expectation that one would avoid heating these lights
unnecessarily. However, the so-called carrier effect o f lights is known to increase the
65
separation o f atmospheric gas-oil from the residue (Golden, 1997). Since steam can act as
a carrier too, the question is whether it can actually substitute for the light hydrocarbons.
energy savings? In other words, would it be worth reducing the yield to achieve energy
efficiency?
All these questions have to be answered taking into account the fact that a specific
quality o f products (given by TBP or ASTM D86 points) needs to be guaranteed equally
the method can be applied directly. In the case o f pre-flashing, the introduction o f the
vapor feed from the flash drum changes the heat load distribution in the column affecting
the heat duty distribution o f the pump around circuits. Thus, to better study the effect o f
the drum temperature and the feed tray, a column without pump around circuits is first
The main column has 34 trays and side products are withdrawn at trays 9, 16 and
25 respectively. The outlet temperature of the furnace is set at 360 °C and the condenser
temperature is fixed at 32 °C. Two crude oils, one light and one heavy are used in the
study. Details o f the assay data and product 95% distillation temperatures were given in
by Bagajewicz and Ji (2000). The over flash rate chosen is 3%, the flowrates o f side
withdrawals and the condenser duty are adjusted to achieve the specifications o f the over
66
flash rate and product 95% distillation temperatures and the flowrates o f stripping steams
are maintained constant. Finally the minimum approach temperature for energy targeting
Consider first a pre-flash drum operated at 163 °C and the feed tray to be tray 15.
Tray 15 is chosen because the vapor concentration best matches the composition o f the
vapor from the drum. The effect o f changing these two values will be studied later. The
simulation results corresponding to a light crude with no pump-around circuits are shown
in Table 3.1 and Figure 3.3. As shown by Bagajewicz (1998), pump-around circuits have
an effect on energy recovery, but a very slight effect on product yield. Thus, as we shall
see later, the energy consumption shown in Table 3.1 should further decrease as pump-
Table 3.1 Comparison between Conventional and Pre-flash Design (Light crude)
Conventional Preflashing
Naphtha, M^/hr 249.6 249.3
Kerosene, M^/hr 144.1 143.9
Diesel, M^/hr 70.5 69.7
Gas oil, M^/hr 118.6 101.9
Residue, M^/hr 212.5 230.6
Product gaps, C
Naphtha-Kerosene 25.1 24.5
Kerosene-Diesel 5 3.2
Diesel-GO 0.6 -0.9
Heating utility, MW 103.6 97.3
Energy consumption, MW 113 106.8
67
10000
V A P O R FEED
^ 8000
i UQUID FEED
O 6000
uT
^ 4000
K
O 2000
^ 1
---------------------------------- ---------- ,-----------
10 20 30 40
TRAY
V A P O R F L O W R A T E :----------------- N O P R E F L A S H ♦ PR E F L A S H
1. The vapor and liquid traffic decrease as a result of decreased heat input in the feed.
2. The yield o f gas oil decreases and the yield o f residue increases as a result o f less
As we shall see later, the difference in yield for the gas-oil and the residues cannot
be reconciled. Essentially, the presence o f lights in the flash zone for the conventional
case, which provide the so-called carrier effect, cannot be substituted by other stripping
means in the case o f the pre-flash or pre-fractionation designs. The comparison with a
68
pre-fractionation design is omitted as the same effect is expected for the reasons just
outlined.
The same comparison was performed for the case o f a heavy crude (Table 3.2). In
Table 3.2 Comparison between Conventional and Pre-flashing Design (heavy crude)
Specifically, the yield of gas oil decreases. As the heavy crude contains less light
distillates, vaporization in the heat exchanger network train is not severe and can be
suppressed under moderate pressure. On the other hand, the heavy crude flashing at a low
temperature (e.g., 163 °C) does not produce much vapor. From the viewpoint o f energy
savings, a small amount o f vapor bypassing the furnace cannot significantly reduce the
duty o f the furnace. We can see from Table 3.2 that the energy consumption for the pre
flash design is only slightly lower. Therefore, the pre-flash design is not justified for
heavy crude. The same can be said for the pre-fractionation design.
69
We now proceed to determine the optimal pump-around circuit loads. The heat is
shifted from the condenser to the pump around circuits step-by-step, as proposed by
Bagajewicz and Ji (2001). Briefly, the method consists o f transferring as much heat as
possible from the condenser to the pump around circuits until the overall minimum
Step 1: Start with a column configuration suggested by Watkins with one exception: No
pump-around circuits.
Step 4: Transfer the optimum amount o f heat to a pump-around circuit located in the
Step 5: If needed, increase the steam in the first side-stripper until the desired gap is
restored.
Step 6: If there is heat surplus from the pump-around circuit just added, transfer the heat
to the next region between draws in the same way as in step 4, if not, stop.
The fact that pre-flashing reduces gas oil yield reveals that the light components
in the crude help the vaporization o f heavy components. However, the yield of gas oil can
also be increased by using more stripping steam. In other words, both light components in
the crude oil and stripping steam have a stripping effect. We now discuss both cases:
70
Table 3.3 Comparison between Conventional and Pre-flash Designs with pump around
circuits (light crude)
Conventional Preflashing
Naphtha, M^/hr 246.6 245.4
Kerosene, M^/hr 143.1 142.6
Diesel, M^/hr 73.9 74.4
Gas oil, M^/hr 119.4 102.4
Residue, M^/hr 212.5 230.6
Product gaps, C
Naphtha-Kerosene 20.5 18.8
Kerosene-Diesel -1.9 -6.6
Diesel-GO -0.3 -1.7
Heating utility, MW 84.2 78.4
Energy consumption, MW 93.7 87.9
In the rest o f the article, we will first explore the carrier effect and its intricacies
in the case of crude fractionation, and then present a comparison between the pre-flash
design and the conventional design for processing the light crude oil on the basis that
For the purpose o f our analysis, the stripping effect o f a component is defined as
the differential change in the amount of residual liquid over the differential change of the
amount of this component in the feed, provided that the system temperature, pressure and
the quantities o f other components are constant. Thus, when the addition of a component
to the feed decreases the liquid rate, the assumption is that it happens at the expense o f
71
vaporizing more o f the heavy or intermediate components. Translated into the crude
fractionation field, the presence of light ends, decreases the residue yield and therefore,
increases the gas oil yield at its expense. We now analyze both stripping agents
separately.
Stichlmair and Fair (1998) present a few charts obtained numerically, showing
that the liquid yield in a flash is lowered when light components are added to a mixture.
We attempt to explain this theoretically. We start using the well-known flash equations:
X: =
(3.1)
where x, and z, are the molar fraction o f component i in the liquid and in the feed, is the
saturated pressure o f component i, P is the total pressure o f the system and L and F are
the liquid flowrate and the feed respectively. The set o f equations (I) is referred to as the
no-steam equation. The vapor-liquid phase behavior is assumed to follow Raoult law,
Let z ' , y ’ be the dry compositions and F ‘ , F ‘ the dry flowrates. The temperature
is assumed to be high enough so that any water in the liquid is neglected. A simple
72
(3.2)
where Pst is the partial pressure o f steam, and P ' is the system pressure. For atmospheric
distillation, P* is somewhere between 0,24 and 0.37 MPa. Since P ' -y, = P^^ x ., we
obtain:
ü P-
(3.3)
X* P -P ..
z.
X i =
+ (1 -
F F
2 ^ x ’, =1 (3.4)
P-
K. =
' P'-P..
We now compare (1) and (4). For a given hydrocarbon mixture flashing at a fixed
temperature, we have z ’ = z,. and F ' = F . Suppose the flash without steam and the flash
with steam take place at the same temperature, then the saturated pressure o f component i
does not change. When P= P ’ - Pst, we have K] = K- and x* = x ,. This means that
injecting steam is equivalent to reducing the system pressure. Note that equation (3) still
holds if steam is replaced by a gas that is insoluble in the hydrocarbon mixture. The use
o f steam, however, has some operational limits that will be explored later.
73
Case 2: Hydrocarbon stripping
We now consider the stripping effect o f light components that distribute in both
I, = ---- t A (3.5)
i = Z ^ (3.6)
j
Because the liquid-phase partition ratio l/fi is less than I, the summation in
equation (7) is less than 1. Accordingly, the denominator is positive. Therefore, whether
the derivative is negative or not depends on the nominator. The nominator consists of two
terms. The first term represents the liquid-phase partition ratio of component / in the
liquid phase, the second term is the average partition ratio o f the mixture. If the partition
ratio o f component i is smaller than the average partition ratio, the derivative is negative,
and the component has stripping effect. Thus, equation (7) can be used to determine if at
a given pressure and temperature, a component is a stripping agent, that is, increasing its
composition in the feed leads to a decrease o f the liquid flowrate in the flash.
studied next. The composition o f a mixture and the K.-values of each component are
shown in Table 3.4. The mixture is allowed to flash at 0.1 MPa and 393 °C. Figure 3.4
74
shows the residue/ feed ratio as a function o f K-values o f the added components. The
molar ratios o f the pure component added to the feed are 1/100, 5/100, 10/100, 20/100
Originally, the residue/feed ratio is 0.43. The residue/feed ratio decreases with an
increasing K-value o f the added component. Note that all curves intersect at K=1.35, at
75
which (— pr equals zero. When the K-value of the added component is greater than
a/, -
1.35, the residue/feed ratio is lower than 0.43. This means these components act as
stripping components. With the increase o f the K-value, the residue/feed ratio decreases.
When the K-value is greater than 30, the residue/feed ratio is almost constant. In this
sense, water, ethane, propane and butane have the same stripping effect on a molar basis.
1.4
1/100 5/100
1.2
o ............. 10/100 X 20/100
I— 1
o — * - — 30/100 — - — - 0/100
0.8 X \
X «
Transition at (1.35, 0.43)
LU 0.6
ID
Q
C/D 0.4 ♦ ♦ ♦ ♦ ♦ ♦♦♦ ♦
LU
CXL \ X X J ................................................... ..
0.2 ^ X x x x x X X X X
0
0.1 1.0 10.0 100.0 1000.0
K
At the same time, it can seen that for a given component with K value greater than
1.35, the residue/feed ratio decreases with an increasing amount o f the added component.
When K-value is less than 1.35, the residue/feed ratio is higher than that without stripping
76
component. This means that the component added is not a stripping agent. Instead, it
prevents vaporization.
vaporization by increasing its partial pressure, but it also hurts vaporization o f other
components by reducing their liquid molar compositions. Furthermore, the part of the
component distributed in the liquid phase will contribute to a larger residue/feed ratio.
When the K value o f a component is large, it generates a large partial pressure, and most
o f it stays in the vapor phase, so the net effect is stripping. On the other hand, when the K
value is small or in other words, the component is heavy, a large portion of the
for crudes, the light crude was mixed with stripping agents and flashed at 360 °C. The
heavy crude was mixed with stripping agents and flashed at 343 °C. The results are
shown in Figure 3.5 and 6, which depict similar trends to that seen in Figure 3.4. The
residue/feed ratio in light crude stripping, however, does not go down as much because o f
the existence of a large amount o f light components in the mixture. The effect is shown
parametric as percentages, which are the amount of stripping agent divided by the amount
In practice, the crude experiences flashing at the flash zone of the main tower, and
the resultant liquid is further stripped with steam at the bottom part of the main tower. To
show how light components affect the flowrate of the residue leaving the main tower, we
split the vapor from the pre-flash drum (163 °C) into two parts as shown in Figure 3.7.
77
One part goes to the main tower and the other mixes with the remaining crude
again and resultant mixture enters the furnace. Now concentrate on the furnace and
examine the effect o f the light components entering into it. The outlet temperature of the
furnace is fixed at 360 °C. By changing the remixing ratio, one can control the amount of
light components entering the furnace. The result is shown in Figure 3.8.
g 0.8
^20%
I 0.6 — 0%
g 0.4
o
CO
g 0.2
0 1 10 100 1000
K-VALUE
ratio. The vapor remixing ratio of one corresponds to the conventional design, and the
ratio o f zero corresponds to the common pre-flash design. It is seen that the yield of the
residue decreases constantly with the increasing vapor remixing ratio. The trend is
78
Following, we first discuss the effect o f steam temperature and the location of its
injection. We tlien investigate the energy requirement associated with the use of
1.2
1. 1 10%
^30%
O 1
UJ 0%
liJ
u_ 0 .9
ÙJ
3
Q
(Ô 0.8
cc
0.7
0.6
0.1 1 10 100 1000
Desalted crude
To flash zone
preflash drum
furnace
79
2 4 0 .0
preflashing
230.0
CO
s
UJ
3
2 conventional
S 220.0
210.0
0 0.2 0.4 0.6 0.8 1
V A PO R SEPARATION RATIO
Figure 3.8 Residue yield as a function of vapor separation ratio in pre-flash design
Typically, there is a temperature drop of about 17 °C between the flash zone and
the bottom tray of the atmospheric column (Watkins, 1979). The temperature drop is
related to the vaporization of relatively light components being directly stripped by the
steam. The question is whether a higher steam temperature reduces the residue yield. As
the atmospheric column does not allow large flowrate o f steam, which would result in the
formation o f free water on top trays, we use the flowchart o f Figure 3.9 to perform this
study.
80
Oil Vapor
Desalted 360 C
Crude
HEN Furnace
Steam
Residue
To show the limit o f the stripping power, the value o f the steam/feed ratio is
allowed to reach high values. Some o f these values are unrealistic, but they help illustrate
the point. When the steam/feed ratio is small, the temperature o f the steam does not affect
the yield of residue because the heat steam extracted from the oil is very small. When the
steam/feed ratio is raised to about 40, the amount of heat taken by the steam becomes
significant, and the 177 °C steam is worse than the 260 °C steam due to a lower
temperature profile at the stripping section of the column. On the 177 °C steam curve, a
minimum point is obser\ ed at a steam/feed ratio o f 75. This means that above this ratio,
the cooling effect o f the steam dominates and the yield o f residue goes up with an
81
360
a I 280
S 2
II 240
200 •
160
0 100 200 300 400 500 600
(a)
240
steam 260 C
200 steam 177C
160 •
I
120
80
0 100 200 300 400 500 600
(b)
Figure 3.10 Residue yield and temperature as functions o f the flowrate o f stripping steam
82
3.6 Steam Injection
In this section, we discuss two cases: injecting steam into the cnide before the
furnace and after the furnace. Figure 3.11 is used for the simulation. The flash zone
pressure is assumed to be the same as that o f the outlet o f the furnace. The results are
Steam/feed, Residue for SBF*, Residue for Flash zone temp, for SAF*, C
KG/M^ M^/hr SAF*, M^/hr
0.0 207.3 207.3 360.0
3.3 202.7 204.1 359.4
5.5 199.8 202.0 359.1
11.0 193.0 197.2 358.3
22.1 181.6 188.6 357.0
*SBF= steam injected at the inlet o f the furnace. SAF = steam injected at the outlet of the furnace.
V apor
Desalted 360 C
Crude
Residue
83
In both cases, the yield o f residue decreases with the increased steam/feed ratio.
For the same steam/feed ratio, however, injecting before the furnace produces less
residue. This is because in this scheme, the steam-crude mixture leaving the furnace has
achieved a vapor-liquid equilibrium and no temperature change takes place between the
outlet of the furnace and the flash zone. In the other case, the hot crude, which is vapor-
liquid mixture at 360 °C, meets with the injected steam (also 360 °C). Because o f the
the temperature o f the system goes down. The last column in Table 3.5 shows the
temperatures in the flash zone. Therefore, injecting steam before the furnace is more
advantageous.
We showed that substances with large K-values have the same stripping effect on
molar basis. However, the energy requirement for heating a gas to an elevated
temperature varies. From the viewpoint o f energy efficiency, gases requiring the lowest
energy are the best. Figure 3.12 shows heating curves for water, Ni, H: and light
hydrocarbons. Water has the largest heating requirement due to the heat o f vaporization.
Nitrogen and hydrogen have the smallest energy demands, and light hydrocarbons lie in
between, with the energy demand increasing with their molecular weight. Therefore,
nitrogen and hydrogen are the best stripping agents. The replacement o f steam by Nz or
Hz has the advantage that corrosion related to water formation in the distillation system is
reduced or even eliminated. The drawback is that they need to be separated from the
84
therefore slightly more complex, and an economic analysis is needed to determine
whether they are economically more beneficial than the conventional steam stripping.
60
50
40
0
30
1 20
10 . -▲
0
0 100 200 300 400 500
TEMPERATURE, C
CH4.g — - — - H2 N2
■C02 ^ ^ "^ ^ w a te r ■ethane, g
H P ro p a n e , g — — N-butane.g
-
The atmospheric tower operates at a pressure in the range o f 0.2-0.3 MPa. Figure
3.13 shows the heating curves for n-butane and N412, which is a pseudo component with
a molecular weight of 167.1, at 0.1 MPa and 0.26 MPa, respectively. The figure shows
that pressure only shifts the temperature at which vaporization takes place. It is also seen
that on a weight basis, the heat requirements for components with large difference in
85
2000
♦ C 4 ,0 .IM P a
X N 4 1 2 ,0 .IM P a *
1500
- - - C4.0.26 MPa
s
N412,0.26 MPa
X
K 1000
M
s
500 X
JKL
100 200 300 400 500
TEMPERATURE, C
hydrocarbon vapor.
The stripping effect o f a hydrocarbon not only depends on its property, but also
The temperature o f the bottom stripping steam does not affect the yield of the residue
Injecting steam before the furnace produces less residue than injecting after the
furnace.
86
• Nz and Hz and light hydrocarbons have the same stripping ability as water but require
less heat to be heated up. From the standpoint o f energy savings, they are better than
water.
• Pressure does not affect the heating curve except postponing the vaporization.
Finally, we can naturally envision two ways to increase the vaporization ratio at a
fixed temperature. One is to put more light components into the crude, which can be
kerosene. The other way is to remove heavy components from the crude. We leave the
conventional option at lower AGO yields. Table 3.6 summarizes the result. In this
comparison, the stripping steams were adjusted so that both designs have the same
shown in Figure 3.14. Not all the heat possible to transfer from the second pump-around
(PA2) to the third (PA3) was transferred because the trade o ff between this load and the
87
Table 3.6 Comparison between Conventional, Pre-flash and Pre-fractionation Designs
(light crude)
Conventional Preflashing Pre fractionation
Naphtha, M^/hr 246.1 245.5 222.5
Kerosene, M^/hr 144.5 144.8 148.8
Diesel, M^/hr 70.7 71.6 71.5
Gas oil, M^/hr 103.6 102.9 105.7
Residue, M^/hr 230.5 230.6 227.5
Product gaps, C
Naphtha-Kerosene 19.9 19.1 18.2
Kerosene-Diesel 0.4 0.0 -0.4
Diesel-GO -2.5 -1.9 -1.7
Heating utility, MW 82.2 76.0 80.4
Energy consumption. 92.5 87.7 90.4
MW
* Pre-flash temperature: 163 °C. Vapor feed at tray 15. AT=22.2 °C.
It is shown that energy consumption o f the pre-flash design is 4.8 MW lower than
that of conventional design. Thus, the energy consumption of the pre-fractionation design
lies between the two designs. To understand the results, two aspects have to be taken into
consideration: one is the temperature level o f the heat to be recovered and another is the
a low-temperature where the heat supply is in surplus, distributing heat between the pre
fractionation condenser and the main column condenser does not affect the heating utility
requirement. Therefore, the only aspect affecting the energy consumption would be the
amount of components bypassing the furnace. When the temperature of the feed entering
the pre-fractionation column is the same as that entering the pre-flash drum, the amount
88
of components bypassing the furnace in the former would be lower because a part of the
vapor entering the column is condensed as the reflux and then mixes with the crude at the
flashing temperature.
Note that product distributions vary considerably from one design to another,
although the product gaps are exactly the same. For example, the pre-flash design gives
more diesel and less gas oil compared to the conventional design. This might be
0.8 PA1
o
0.6 COND CRUDE
CL 0.4
o PA3
*
0.2 PA2
RES
TEMPERATURE, 0
Figure 3.14 Final Heat Supply-demand Diagram for a Pre-flash Design (AT= 22.2 °C)
89
3.10 Effect of pre-flash tem perature
The necessity of a pre-flash drum depends on the property o f the crude, the
temperature o f the crude entering the furnace and the operating pressure. The bubble
temperatures o f the lighter crude are higher than that of the heavier crude (Figure 3.15).
The higher the operating pressure, the higher is the bubble temperature. If the
temperature o f a crude oil exceeds its bubble temperature considerably before entering
300
250
200
UJ
Q.
S A P I= 3 6 .0
UJ
»- 150
A P I= 3 2 .8
100
0 0.5 1 1.5 2
PRESSURE, MPa
The effect of the pre-flash drum temperature on the low AGO yield arrangement
is shown in Figure 3.16. With the increase o f the pre-flash temperature, the heating
utility decreases continuously because o f more light components bypassing the furnace.
90
On the other hand, the steam consumption increases rapidly as more steams are needed
for both keeping the gas oil yield and fixing the product gaps. Although the total energy
consumption reaches a minimum at 177 °C, the differences in the curve are relatively
small. However, if steam is in surplus in the refinery, heating utility only counts and
110
h 50
Total energy consum ption
100
- 40
S
O 90 H eating utility - 30
a:
ill
Z
UJ 20
80 Steam consum ption
- 10
70
160 180 200 220
TEMPERATURE, C
The vapor produced in the pre-flash drum at 163 °C has properties very similar to
naphtha. It contains 53% light ends (C 2 -C 5 ) Its 98% boiling point is 169 °C, lower than
91
the 95% temperature o f naphtha (182 °C). We compared the effect of locating this feed
in different trays in the column. During the comparison steam flowrates have been kept
constant, allowing the gaps to change. The results are given in Table 3.7.
1. The naphtha yield and the gap between naphtha and kerosene are constant. This is
because the vapor feed is basically naphtha components and the naphtha-kerosene
separation section (tray 1 to tray 9) is almost not affected in the above feeding
locations.
2. When the vapor feed location is lower than that o f the gas oil withdrawal (tray 25),
the yield o f diesel is significantly lower and the yield o f gas oil is higher. This is
92
because that the temperature of vapor feed is much lower than that o f the vapor rising
from the flash zone and a portion o f the hot vapor is condensed with the effect of
some diesel components going to the gas oil draw stream. In Table 3.7, we can see
when the vapor feed tray is 27, not only the yield o f diesel decreases but also the
diesel-gas oil gap moves in the negative direction. Therefore, vapor feed between
3. When the vapor feed is sent to a tray above the flash zone, the residue yield is
constant. This is because that the residue yield depends on four factors: the
composition and temperature o f the crude at the outlet of the furnace, the flowrate of
stripping steam entering from the bottom o f the main column, and the over flash rate.
In the above situations, all the four factors are the same.
380
-T R A Y 8
18
360 - tra y
Ü A TRAY 27
UJ X TRAY 29
a: 340
z>
:
LU
CL 320
2
LU
h-
300
280
15 17 19 21 23 25 27 29 31 33
TR AY
93
The temperature profiles are shown in Figure 3.17. The tray temperatures are
highest when the vapor is fed at tray 8, the highest location. When the vapor is fed at tray
18, the tray temperatures above the vapor-feeding tray are lower, which is due to the
cooling effect o f the low temperature-feeding vapor. The tray temperatures below the
vapor feed tray are exactly the same as that when the vapor is fed at tray 8. When the
vapor feed location continues to move down, the same tendency appears as expected.
5000
i
—I TRAY8
0
s 4000 + TRAY 27
§ ^ ^ A A TRAY29
3000
1
“h
2000
▲
1000
+ ■A’ -4 r -4 r T
0
24 26 28 30 32 34
TRAY
Finally, when the vapor location reaches its lowest location — the flash zone (tray
29), the temperatures o f tray 29 through tray 34 are much lower than all the others with
other vapor feed location. At the flash zone, the hot crude flashes and splits into rising hot
vapor and descending hot unstripped residue. The hot vapor meets with the cold pre-
94
flashed vapor immediately and condensation occurs. The condensed liquid, basically gas
oil and diesel components, mixes with the unstripped residue on the tray 29. However,
the condensed liquid does not stay with the unstripped residue and is stripped out when
the mixture of the unstripped residue and the condensate meets the steam coming from
tray 30.
In Figure 3.18, we can see that the vapor flowrate o f tray 29 is about twice as
large as that when the pre-flashed vapor is fed at other locations. The overall effect is
equivalent to that o f conventional design with lower outlet temperature of furnace. It can
be predicted that heating o f the pre-flashed vapor to higher temperature would reduce the
From the above comparison, we can conclude that product yields and gaps are not
sensitive to vapor feed location. Specifically, the residue yield keeps constant. As a
higher vapor feed is preferred. However, as pointed by Golden (1997), a higher vapor
feed would have the risk o f contaminating the products withdrawn below the feed tray
3.12 Conclusion
A rigorous targeting for the design of a crude distillation unit with pre-flash
drums and pre-fractionation is presented. The major conclusion o f this study is that pre-
flashing and pre-fractionation designs can only be advantageous from the point of view
o f energy consumption if one accepts the yield of gas oil to be reduced. Other findings of
95
the study show that in practice, the equivalent o f the carrier effect o f lights cannot always
advantageous to pay the penalty o f higher energy consumption just to see the number o f
units in the preheating train reduced. Finally, heat efficiency and network simplicity is
desired for a variety o f crudes simultaneously. This, together with measures to improve
3.13 Nomenclature
96
X, = molar fraction o f component i in the liquid phase
3.14 References
1. Andrecovich, M., and Westerberg, A., A Simple Synthesis Method Based on Utility
(1985).
Atmospheric Crude Fractionation Units Part II: Heat Exchanger Networks. Industrial
5. Golden, S., Prevent Pre-flash Drum Foaming, Hydrocarbon Processing, May 1997,
ppl41-153.
6. Stichlmair, J.G, and Fair J. R, Distillation Principles and Practices, Wiley-VCH, New
York, 1998.p76.
97
7. Terranova, B., and Westerberg, A., Temperature-Heat Diagrams for Complex
1374-1379,(1989).
98
Chapter 4 On the Energy Efficiency of Stripping-Type Crude
Distillation
4.1 Overview
columns is analyzed. Rigorous simulations prove that in the case o f processing a light
crude of reference, previous reports indicating that this option is advantageous over the
4.2 Introduction
In conventional crude atmospheric distillation, the crude is fed at the bottom part
of the column, and the main part of the column functions as a rectifying section. Watkins
(1979) suggested the possibility of using direct and indirect rectifying sequences. The
conventional design that he chose to describe in detail is shown in Figure 4.1. Its
years, the alternative indirect sequence (Figure 4.3) was proposed by Liebmann (1995).
In previous work (Bagajewicz and Ji, 2001, Bagajewicz and Soto, 2001), a
systematic procedure for the design o f conventional crude fractionation units was
and heat integration. The procedure starts with a column without pump around circuits.
When heat is transferred from the condenser to pump around circuits, a trade off between
the steam usage and the fuel gas savings is established. This transfer of heat is possible
due to the operating and design flexibility that crude fractionation installations exhibit.
99
Such design flexibility was studied in detail by Bagajewicz (1998). The procedure makes
use o f rigorous simulations and heat supply-demand diagrams similar to those introduced
by Andrecovich and Westerberg (1985), and Terranova and Westerberg (1989). Based on
these targets, a universal heat exchanger network which allows several crudes to be
processed with the minimal energy consumption can be found (Bagajewicz and Soto,
2001 ).
^CONDENSER
~\J~ ^wjiter
> naphtha
PAl
steam
PA2 kerosene
steam
PA3 diesel
water — T
crude— DESALTER steam
HEN g asoil
sour water!
— I steam
'
V
HEN FURNACE >■ residue
100
Condenser
-► Naphtha
Kerosene
-► Diesel
Residue
(about 150°C) and fed at the top o f the column. Because the crude temperature is low, the
vapor ratio of the feed is small. The crude goes down the column and is heated
consecutively in three heaters (the upper heater, middle heater and lower heater). Side
products are withdrawn from the vapor phases, and rectified in the side rectifiers.
Liebmann and Dohle (1995) reported that this design could feature 5% less utility
cost than the optimized conventional design. However, the comparison was not on the
same allowable temperature basis. The maximum allowable temperature is limited by the
thermal stability property o f the crude being processed and is found by lab testing. For
101
the Venezuela crude oil used in Liebmann’s paper, the allowable temperature is 343 °C
(Watkins, 1979). However, in their stripping-type design, the crude was heated to 370 °C
in both the middle heater and the lower heater. It can be expected that at this temperature,
severe thermal cracking takes place. Such operation is not allowed in practice. Because o f
necessary. This chapter performs the evaluation using the method proposed by
Bagajewicz and Ji (2001), taking into account the temperature limit of thermal cracking.
Preheater ^ Water
Crude oil
j^. -» Naphtha
”0 Kerosene
UH
-► D iesel
MH
"► Gas oil
LH
Steam
Residue
102
4.3 C rude vaporization patterns and beat dem and
A major difference between the conventional design and the stripping-type design
is the heating pattern o f the crude oil. Because in the stripping-type design crude oil is
heated step-by-step and the vapor is separated immediately after it is generated, the
amount of crude to be heated in the stripping section is smaller than in the conventional
design. On the contrary, the amount of crude being heated in the conventional design is
constant, that is, no vapor is separated until the heating is completed at the outlet o f the
furnace.
To better understand the difference between the two heating patterns, one should
analyze the distillation curves. It is well known that distillation curves can be used to
ratio. The two relevant distillation curves, ASTM D86 and EFV, are useful tools.
ASTM D86 distillation is carried out in a glass flask equipped with an electric
heater and a water cooler. The oil sample is heated in the flask and the resultant vapor is
ratios of crude oils as a function of temperature. In this test, the crude oil is continuously
heated without separating the vapor from the remaining liquid (Nelson, 1958). The
temperature versus liquid amount is recorded. To obtain the EFV curve, a series o f runs
at different temperatures are earned out, and each run constitutes one point (of
103
The equilibrium flash vaporization (EFV) distillation provides a relation between
system temperature and the percentage o f vapor generated. EFV distillation is a batch
process during which no vapor is separated. In principle, the heating process in EFV is
similar to that in the conventional distillation. In practice, the EFV curve is used to
estimate the vaporization ratio for the conventional design (Nelson, 1958).
Similarly, a close relation between ASTM D86 distillation and the stripping-type
with a vapor withdraw line between adjacent heaters as well as zero pressure drop in the
tower, one would obtain exactly the same temperature vs. vaporization curve as ASTM
D86 distillation.
Now the ASTM D86 and EFV curves are used to answer two questions:
• For the same temperature limit, which distillation option produces more vapor, this is,
• To achieve the same vaporization ratio, which distillation option demands more heat?
The first question can also be rephrased as follows: to achieve the same
maximal temperature. Although the answer to this is well-known, a review will highlight
The ASTM curve is located above the EFV curve. This means that at the same
temperature limit, ASTM distillation always produces less vapor than EFV distillation. In
this sense, the stripping-type design will likely produce less amount of distillates and
104
more residue under the same temperature limit. Whether other operating variables, such
500
Ü 400 A STM
lif EFV
K
g 300
u 200
Q.
s
UJ
I- 100
Figure 4.4 ASTM curve and EFV curve o f the Venezuela crude (see Appendix A)
From the viewpoint of energy efficiency, the stripping-type design reduces the heat
demand from the light components (naphtha and kerosene components), which are only
heated to low temperatures. Conversely, in the conventional design, light components are
heated to the temperature limit. The heating pattern for the stripping-type design (ASTM)
is, however, not advantageous for the heavy components (gas oil components). As
mentioned before, the distillation temperature for the ASTM to achieve the same
vaporization ratio is higher. This means that the heavy components vaporize at a higher
105
temperature than they would in the EFV distillation. Therefore, the heat demand for
WTiether ASTM requires less energy to achieve the same vaporization ratio,
ignoring the temperature limit o f thermal cracking, will depend on the trade-off between
the heat saving for light components and the larger energy consumption for heavy
components. To obtain the ASTM heating curve, 9 heaters and 9 separators were used to
simulate the ASTM distillation. Figure 4.5 shows part o f the flow sheet used for this
simulation. The crude was heated at 760 mmHg step-by-step and vapors were withdrawn
in each step. The resulting curves for heat demand are shown in Figure 4.6.
VI
—►
V3
Heater HI
Heater I Heater II L2 T3
Heaters 4 through 9 and separators 4 through 9 are not shown in this figure.
Figure 4.6 shows that for small vaporization ratios (less than 0.25), the two curves
coincide. For larger vaporization ratios, however, the heat demand for ASTM is always
106
higher than that for EFV distillation. To explain this, a comparison of temperatures and
heat demands needed for achieving 0.473 vaporization are listed in Table 4.1.
Because the vapor generated is not separated in EFV distillation, on average 24%
o f the crude in the form o f vapor is overheated from 100 °C to 307 °C. In ASTM
distillation, the overheating o f vapor is entirely avoided, as the vapor is withdrawn from
the system after it is generated. However, ASTM distillation has to further heat the liquid
(59% of the crude) from 307 °C to 380 °C. The above analysis means that the effort to
reduce the heat demand by avoiding overheating o f light components in the stripping-
1000
Z — ASTM
800
— EFV
600
<
s
UJ 400
Q
< 200
UJ
z
0
0 0.2 0.4 0.6
VAPOR FRACTION, WT
Figure 4.6 Heat requirements for ASTM distillation and EFV distillation at 760
107
Table 4.1. Comparison of temperature and heat demand
As discussed in the previous section, the stripping-type design has to heat the
the conventional design. In addition, liglit components help vaporize heavy components
at lower temperatures. This is called the carrier effect and it was discussed in detail by Ji
and Bagajewicz (2001). Such light components are not present in the stripping-type
design.
conventional design, all light components come from the bottom section of the tower and
go through the trays where side products are withdrawn. As a result, light components
appear in each side product withdraw line and have to be stripped off in the side strippers.
In industry, this is called controlling the flash point. In contrast, the light components in
the stripping-type design come from the top section and are withdrawn as soon as they
are vaporized, not reaching therefore the trays where diesel and gas oil are withdrawn.
Thus, the problem o f the presence of light components does not exist. However, heavy
108
components have a large chance of being present in these side products and raise their
end points.
The last feature is that the products are withdrawn in vapor phase instead o f liquid
phase. For the same composition of mixture, the dew point is higher than the boiling
point. So, by withdrawing a vapor one may obtain the heat at a higher temperature level.
This is advantageous from the point o f view o f energy efficiency. However, the vapor
phase withdrawals also bring the corrosion problem to the side condensers if water is
present.
The main column in Figure 4.1 contains 34 trays. A crude o f API 36 is used in the
simulation. The flowrate used is 5000 BBL/hr. The crude data were taken from a
previous chapter (Bagajewicz and Ji, 2001). The properties of the crude are listed in
109
In simulating the stripping-type distillation, the feed temperature, the duties of
upper, middle and low heaters, the duties o f condensers are allowed to change in order to
meet the specifications. Comparisons are made with a conventional column operating at
Figure 4.7 is a heat demand-supply diagram for the stripping-type design. The
solid bold line stands for the crude heat demand curve, which is composed of the
preheating section and the upper, middle and lower side heaters. The heat supply includes
heat fi"om the main condenser, side condensers, products and the residue. The demand
curve is not continuous due to temperature gaps between adjacent heat sinks.
The discontinuity o f the heat demand curve is an important feature. It affects the
match between the heat demand and the heat supply. Consider the region between the
preheating curve and the upper heater. In this region, heat supply is available from side
condensers and the residue. However, there is no heat demand in this region. The only
way to make use o f this heat supply is to cascade the heat to regions at lower
temperatures. In the preheating region, a heat deficit is present (shaded region), but it is
much smaller than the heat supply aforementioned. Therefore, a large part of the heat
supply can not be utilized. Finally a small match between supply and demand exists at the
level o f UH.
110
1.6
Pinch
MH
2
UH
i Heat deficit
0.8
a SC1
u Condenser Crude
SC3
LH
.4 SC2
Residue
SW
Res.
Kerosene
0
0 100 200 300 400
TEMPERATURE, C
III
The location o f the pinch point can be easily obtained from this diagram. It is the
lowest temperature at which the demand is larger than the supply after the shifting and
area matching has been performed (Figure 4.7). The heating utility is given by the
unmatched demand on the right, and the cooling utility is given by the extra supply on the
left.
The heating utility can be reduced by raising the temperature o f the feed. That is,
by shifting part o f the upper heater duty into the crude preheater. The solid curve in
Figure 4.8 represents the new heat demand curve for a higher feed temperature and the
dashed line corresponds to the basis case. Figure 4.9 shows the complete diagram for the
1.6
Pinch MH
1.2
New demand UH
curve
0.8
u«
LH
0.4
0
0 100 200 300 400
TEMPERATURE, C
112
1.6
MH
I- 0.8
SCI
UH
S Condenser Crude
SC3 LH
0 .4 SC2
Residue
SW
Res.
Kerosene
Figure 4.9 The complete heat demand-supply diagram for the higher feed temperature
113
Note that the location o f the pinch does not change at all, however the heat deficit
in the upper heater region decreases significantly. The demand of the middle and lower
A comparison o f the strategy used above with that for the conventional design
reveals a few important issues. With a fixed heat demand curve (the crude curve) in the
conventional design, the strategy is to move the extra heat supply to higher temperature
regions, which was realized by redistributing the pump-around duties (Bagajewicz and Ji,
2001). In the heat demand curve in the stripping-type design can be modified by
redistributing heat among the preheater and the side heaters. Here the strategy is to move
some heat demand from the high temperature region (the upper heater region) to the low
However, there is a limit on the heat shift from the upper heater to the preheater.
Table 4.3 shows that the naphtha-kerosene gap decreases with the increase o f the feed
temperature. This is because the vapor arising fi’o m the upper heater decreases in amount
as a result o f a lower upper heater duty. In the section o f the column between the feed
tray and the upper heater, the vapor strips off light components from the descending
crude.
When the vapor flowrate is too low, the vapor cannot efficiently remove the light
components from the crude, and the crude entering the upper heater will carry a
significant amount o f light components. In the upper heater, these components vaporize
along with kerosene and enter the kerosene rectifier. As the rectifier has no way to
remove light components, all the light components are present in the kerosene product,
114
Table 4.3 Effect o f Feed Temperature in Stripping-type Design
237.8 °C 260 °C
Preheater duty, MW 60.0 72.4
Upper heater duty, MW 42.3 31.9
Naphtha yield, M^/hr 244.8 235.2
Kerosene yield, M^/hr 141.0 155.1
Naphtha-kerosene gap, °C 18.8 5.3
The residue yield o f the stripping-type design is about 70% larger than that o f the
conventional design for the same temperature limit o f 360 °C. This yield cannot be
temperature and/or the duties o f the heaters. Because the residue yield has significant
was run at this large yield. As a result, the maximum temperature in the conventional
column drops to 324 °C. While the yields o f naphtha are almost the same, however, the
yield o f kerosene is significantly higher, and the yield o f diesel is lower in the stripping-
type design (Table 4.4). This is because in the conventional design, the light components,
which constitute the kerosene, ascend through the trays where the gas oil and the diesel
are withdrawn.
Some o f these light components are carried out with the diesel and the gas oil. In
the stripping-type design, the crude oil enters the column at the top. Therefore, the
carrying o f kerosene in the heavy products is ruled out. As expected, in the stripping-type
design the gap between kerosene and diesel is much higher than that in the conventional
design. Finally, the energy consumption is calculated: adding the minimum heating utility
115
o f the heat exchanger network and 70% o f the steam consumption, to account for cost
Conventional Stripping-type
Naphtha-Kerosene Gap (°C) 16.7 16.7
Kerosene-Diesel Gap (°C) 0.0 13.6
Diesel-Gas oil Gap (°C) -13.2 -13.9
Naphtha yield, M^/hr 244.7 242.1
Kerosene yield, M^/hr 137.8 157.4
Diesel yield, M^/hr 52.8 35.9
Gas oil )ield, M^/hr 43.9 35.0
Residue yield, M^/hr 316.4 324.5
Heating utility (furnaces, MW) 45.6 49.5
Steam consumption (MW) 19.1 16.4
Energy consumption (MW) 59.0 61.0
temperature limit (399 °C) to obtain the same yield o f residue as in the conventional
column (360 °C). It is at such higher temperatures, which are not recommended in
practice, when the stripping type design achieves around 6% less of total energy
consumption.
Investment cost
The above comparison shows the stripping-type design can not achieve the same
distillates yields as the conventional design for the same allowable heating temperature.
116
Since the crude is heated to a lower temperature, a less complex preheat train may be
needed. Thus the trade-off between operating and investment costs needs to be analyzed.
The relative costs of both designs are shown in Table 4.5. The main tower for the
stripping-type design is considered to be higher because the dirty crude goes through
most o f the trays. The dirty material reduces the tray efficiency and causes fouling
problems. Therefore, additional trays are needed to offset the lower efficiency and
especial designs are required to reduce fouling. Both requirements contribute to a higher
The side condensers used in the stripping-type design are considered more
expensive than pump-around heat exchangers in the conventional design because the side
117
condensers have to deal with corrosion problem caused by the steam condensation. The
corrosion problem becomes more severe when processing high sulfur crudes.
The investment cost for the furnace in the stripping-type design is considered
higher because it handles four cold streams. The internal structure will be more complex
than that of the conventional design where only one cold stream exists. Besides the
The relative cost for all the other heat exchangers is not easy to estimate without
designing the heat exchanger network. However, a comparison can be made on the bases
of total number o f exchangers because it is the dominating factor in the total cost. The
For the case o f the conventional colunm in the region above the desalter (Figure
4.10), there is one cold stream and 7 hot streams: P A l, PA2, PA3, residue, gas oil, diesel
and kerosene.
By using the (N-1) rule (Biegler et al, 1997), seven heat exchangers and a heater
are needed. The condenser is not included because the heat is in surplus and is cascaded
into the region below the desalter. In the region below the desalter, there is a large heat
surplus. Since the heat from the condenser is large enough, only one exchanger is needed.
In addition, the four distillates and the condenser have to be cooled to the final
For the stripping type design above 260 °C, 3 exchangers are needed for the crude
to extract heat from SC2, SC3 and the residue (Figure 4.11).
118
1.4
^ 0 .6 COND
119
UH
I
s 0.8
SC1
MH
COND
Ô 0-6 SC3 LH
SC2
RESIDUE
0.2
KEROSENE
Distillation
SW: saline water. SCI: side condenser 1. SC2: side condenser 2. SC3: side
condenser 3.
120
The heat supply to the left o f UH has to be cascaded to lower temperatures for
utilization. Above the desalter and below 260 °C, there are 5 hot streams: the residue,
SC 1, kerosene, diesel and gas oil. Similar to the conventional design, the condenser is not
used. The number o f exchangers is 5 in addition to the heater. Below the desalter, one
exchanger is counted for heat exchange between the condenser and the crude. Four
coolers are required for cooling the products. The total number o f the exchangers besides
Both designs take the same number o f heat exchangers. Assume the number of
exchangers is the dominating factor in pricing, we expect the costs for the exchangers to
be close to each other. It is then concluded that the stripping-type design requires higher
4.7 Conclusions
From this study one can conclude that the stripping-type design for crude
fractionation cannot achieve low yields o f residue, and under the same yield of products,
it is not as efficient as previously reported. The difference may rely on different crudes
used as well as design procedures that utilize temperatures that are too high.
121
4.8 References
3. Ji Shuncheng and Bagajewicz, M., Rigorous targeting procedure for the design of
4. Liebmann, K., Integrated crude oil distillation design, Ph.D. dissertation. University
5. Liebmann, K., and Dhole, V. R., Integrated Crude Distillation Design. Computers &
Crude Oil Distillation Tower Using Pinch Analysis. Inst, o f Chem. Engineers, 76(3),
7. Glinos, K. and Malone, M. P., 1985, Minimum vapor flows in a distillation column
with a side-stream stripper. Ind. Eng. Chem. Proc. Des. Dev. 24(4); 1087-1090.
1958.
9. Sharma, R.; Jindal, A.; Mandawala, D.; Jana, S. K. Design/Retrofit targets of Pump-
around Refluxes for Better Energy Integration o f a Crude Distillation Column. Ind.
122
10. Watkins, R.N., Petroleum Refinery Distillation, 2""^ edition. Gulf Publishers, 1979.
4.9 Appendix
Appendix A
The properties o f the light crude are presented. There are taken from Watkins
(1979). The crude is TIA JUANA light (Venezuela) having an API gravity o f 31.6.
123
Appendix B
The properties o f the crude (API 36) used in Table 4.3 and 4.4 are shown here.
Appendix C
(1996). The designs were carried out using ASPEN plus release 8.5-6. Column 1 is the
124
stripping-type design. Column 2 is the conventional design. The conventional design was
simulated in this study using PRO II 5.11 and the results are listed in the last column.
125
Chapter 5 Design of Crude Distillation plants with vacuum
5.1 Overview
In previous work, the targeting procedures for atmospheric distillation plants were
presented. These procedures were developed assuming that the atmospheric residue
leaves the plant as a product. In many cases, however, this residue is distilled in vacuum
towers to further extract more gas-oil. The addition of a vacuum tower has many
the addition o f a vacuum column favors the pre-flashing distillation arrangement, which
5.2 Introduction
vacuum distillation. Despite o f the importance o f crude distillation, studies on its optimal
design are scarce. One o f the important early publications is “Petroleum Refinery
Engineering” by Nelson (1936). This book discusses the crude distillation in detail. It
contains many charts and experimental data. The 4*'’ edition was published in 1958. In the
design procedure presented in this book, the amount o f products is estimated from the
crude distillation curve, then the number o f trays between side withdraws are picked.
Next, the temperature and pressure in each tray is calculated using empirical charts. Heat
126
balance is then carried out and the reflux is determined. The design procedure ends with
Much later, another book on crude distillation was published by Watkins (1979).
This procedure provided in this book is similar to the one proposed by Nelson (1958). In
this design procedure, the distillation tower is designed first and followed by the design
of heat exchanger network. With this procedure, it is difficult to take into account the
interaction between the crude tower design and the heat exchanger network. Liebmann
(1998) proposed an integrated design procedure for the design o f the atmospheric tower.
The design procedure starts with a sequence o f simple columns that are generated by
decomposing the crude main tower. Next, reboilers and thermal coupling are introduced
in order to reduce utility consumption. The grand composite curve ( a tool used in Pinch
Technology) is used to assess the proposed modifications. After all the possible design
modifications have been explored, these columns are merged into a single complex
column.
around heat removal. First, a practical minimum reflux ratio for each column section is
determined using Packie’s empirical diagram (Packie, 1941). Then, the heat removal in
the upper part o f the column is calculated using a heat balance. The upper part may start
from an arbitrary tray and end with the condenser. Following, the upper part is extended
tray by tray and heat surplus is calculated for each tray. The resultant heat surplus data
are used to construct a column grand composite curve. Finally, the maximum heat
removal for each section is determined using the column grand composite curve. A major
advantage o f this method is that the maximum heat removal can be estimated quickly
127
wiü-.out the need o f simulation. However, as Packie’s diagram is empirical and the effect
o f the stripping steam is not included, the heat removal calculated is not accurate. This
suggests a procedure based on rigorous simulations that can capture the relationships
between the column variables and the heat integration opportunities accurately.
Bagajewicz (1998) studied the flexibility aspects for the design o f atmospheric
columns using rigorous simulators. Bagajewicz and Ji (2001) proposed such a rigorous
targeting design procedure has been proposed for the design o f conventional crude
distillation units. This procedure aims at flnding the best scheme for a multipurpose crude
distillation unit that processes a variety o f crudes, which is an issue that had been
grand composite curves are used as a guide directing the search for optimal schemes. An
advantage o f heat demand-supply diagrams is that the role o f each stream, heater or
cooler in the total energy consumption is clearly shown, so the search of the best scheme
Following the above procedure, the energy targets for a light crude and a heavy
crude were obtained. The targets were later used for the design of a multipurpose heat
exchanger network (Bagajewicz and Soto, 2001). The model for the multi-purpose heat
exchanger network takes advantage of the flexibility identified and is able to predict the
desalter temperature that needs to be used. A conjecture that a design for extreme crudes
would be able to accommodate the processing o f intermediate crude was posed and
partially confirmed.
128
An important variant o f atmospheric distillation is pre-flashing or pre
fractionation design. The pre-flashing and pre-fractionation designs are put in the same
category because both feature vapor separation before the furnace. Ji and Bagajewicz
(2001) presented a systematic procedure to obtain design targets for heat integration in
drums. It is shown that under the same high product yield conditions, pre-fractionation or
pre-flashing are not advantageous from the energy point of view. This is in great part due
to the loss o f the carrier effect that light components have in separating heavy gas-oil
fractions in the flash zone. The study also showed that the equivalent o f the carrier effect
Soto and Bagajewicz (2001) noted that extensive splitting o f crude streams has
several drawbacks. For example, not much energy efficiency is lost if the branching is
reduced and therefore the trade-off between heat utility and capital cost breaks even. The
model proposed Soto and Bagajewicz (2001) by uses binary variables to account for the
processing o f different crudes and count heat exchangers. The second part o f this paper
improves the model by allowing a match between a given pair to take place in different
intervals in different periods. Finally, none o f recent papers addressed the heat exchanger
network design for crude distillation with either the vacuum tower or with preflashing.
and vacuum distillation are first studied. Next, targeting procedures for all alternatives
design are proposed. These procedures are used to find targets for the light crude. Finally,
129
5.3 Alternatives for combined atmospheric-vacuum distillation
Figure 5.1 shows the conventional design of an atmospheric unit. Crude is mixed
with water and heated in a heat exchanger network before entering a desalter where most
o f the water containing the salt is removed. The desalted crude enters another heat
exchanger network and receives heat from hot streams. Both heat exchanger networks
make use o f the vapors o f the main column condenser, the pump-around circuit streams,
and the products that need to be cooled. The preheated crude then enters the furnace,
where it is heated to about 340-370 °C. The partially vaporized crude is fed into the flash
zone o f the atmospheric column, where the vapor and liquid separate. The vapor includes
all the components that comprise the products, while the liquid is the residue with a small
amount o f components in the range o f gas oil. These components are removed from the
p . w ater
*■ n aphtha
PA l
PA 2 kerosene
PA 3 diesel
w ater D ESA LTER
C ru d e
HEN
gas oil
steam
so u r w a te r HEN FU R N A C E
R e sid u e lo Vac. T ow er
130
In addition to the overhead condenser, there are several pump-around circuits
along the column, where liquid streams are withdrawn, cooled, and sent back to upper
trays. Such arrangement distributes the reflux throughout the column. It also has
beneficial effects for heat integration because the temperature o f these liquid draws is
higher than that of the condenser. Products are withdrawn in liquid state from different
trays and then stripped by steam in side strippers to remove light components. The
residue from the main tower is heated further in the vacuum furnace and sent to the
condenser
naphtha
PAl ^
S te a m
PA2 ^ kerosene
Crude Oil
vapor
D esalter steam
PA3 ^
diesel
steam
sour w ater gas oil
Pre flash drum steam
Furnace AR to Vac. Tower
131
In the preheat train, the crude is under pressure in order to suppress vaporization.
In some cases, however, the crude is so light that above certain temperature, the pressure
required to suppress vaporization is too high. The solution is to separate some light
components before heating the crude further in the preheat train. The light components
A systematic procedure was presented (Ji and Bagajewicz, 2001) to obtain design
targets for heat integration in conventional crude fractionation units that use pre
fractionation columns or pre-flash drums. It is shown that under the usual temperature
limits to prevent thermal cracking, the pre-flash design generates more atmospheric
residue and less gas oil. In a complete plant, the atmospheric gas oil loss can be picked up
by the vacuum tower. The price is that the vacuum tower has to handle a larger amount o f
feed stock and the vacuum jet steam consumption increases. Whether the trade-off is
In the stripping-type design (Figure 5.3), the crude is heated to a relative low
temperature (about 150°C) and fed at the top o f the column. Because the crude
temperature is low, tlie vapor ratio o f the feed is small. The crude goes down the column
and is heated consecutively in three heaters (the upper heater, middle heater and lower
heater). Side products are withdrawn from the vapor phases, and rectified in side
rectifiers.
Liebmann and Dohle (1995) reported that for atmospheric distillation, the
stripping-type design could feature 5% less utility cost than the optimized conventional
design (Figure 5.1). However, the comparison is not appropriate because in the stripping-
type design, the crude oil was heated to a much higher temperature than in the
132
conventional design. The maximum allowable temperature is limited by the thermal
stability property o f the crude being processed and is found by lab testing. For the
Venezuela crude oil used in Liebmann’s paper, the allowable temperature is 343 °C
the crude was heated to 370 °C in both the middle heater and the lower heater (see
Appendix B). It can be expected that at this temperature, severe thermal cracking will
take place. Such operation is not allowed in practice. Because o f the above limitation of
(2001) took into account the temperature limit o f thermal cracking and made such
evaluation.
Preheater Water
> Naphtha
Desalted crude Kerosene
UH
^ Diesel
MH
► Gas oil
LH
Steam
>■ AR to Vac. Tower
133
Similar to the preflash design, the loss o f atmospheric gas oil is picked up in the
vacuum column at the expense of extra vacuum jet steam consumption. In this paper, the
question of whether the total energy consumption is lower than in the conventional
design is evaluated.
In this section features regarding vacuum distillation that are relevant for the
The topped crude leaving the atmospheric tower still contains significant amount
pressure because the temperature required would be so high that severe thermal cracking
takes place. Vacuum distillation is based on the principle that the boiling point of a
whole crude cut points between distillates and the atmospheric residual are somewhere
from 370 to 430 °C. The cut point can be raised to 580 to 610 °C in vacuum distillation.
Thus, the general function of the vacuum tower is to remove the maximum possible
amount o f distillate from the charge stock while meeting product specifications (Watkins,
1979).
Figure 5.4 is a schematic diagram for production of light vacuum gas oil (LVGO)
and heavy vacuum gas oil (HVGO), but sometimes, depending on its properties, LVGO
is blended with other products like diesel. Both are typically used as feed to fluid
catalytic cracking units. The vacuum distillation consists o f the vacuum furnace, vacuum
tower and the vacuum producing system. The topped crude is heated up in the vacuum
134
furnace to about 400 °C. The temperature is controlled to be ju st below the temperature
of thermal decomposition. The heated oil partially vaporizes in the furnace and the vapor-
liquid mixture enters the flash zone of the vacuum tower. The vapor ascends in the
column and condenses in the HVGO pump-around section and the LVGO pump-around
section. Note that no strippers are used for the products because the initial boiling points
♦ To Ejectors
♦ VLGO
» VHGO
Topped Crude
Vacuum Furnace
Vacuum Residue
The HVGO section is the main heat removal zone. HVGO is withdrawn and
cooled down in the heat exchanger network and then a portion o f HVGO returns to the
top o f the packing. Although a single cut o f VGO (vacuum gas oil) is allowed in some
cases, drawing LVGO and HVGO separately is more beneficial from the point o f view o f
135
energy savings, because the resultant HVGO draw temperature is 90-120 °C higher than
The LVGO section is the top zone in the tower where the LVGO is condensed
and separated from the non-condensables, cooled and a portion circulated to the top o f
The temperature o f the steam and non-condensable materials leaving the top o f
practical cool oil temperature in the top pumparound circuit. The latter temperature is a
function o f the viscosity properties o f the oil processed. Usually a cooling oil temperature
The residue section serves two purposes: One is to remove relatively light
components, the other is to reduce coke formation. The former is achieved by steam
stripping using a steam to net residue ratio o f 4-6 Ib/bbl. The latter is realized by
circulating the partially cooled bottoms to quench the liquid to 365 °C.
The non-condensable gases and steam leave the tower from the top and are
ejected from the system by means of the steam jets, thus creating the required vacuum.
3. Air dissolved in the feedstock and in the water used in generating steam.
4. Air leakage.
136
The steam consumption consists o f two parts: process steam and vacuum jet
steam. The total steam consumption depends on the absolute pressure in the vapor line,
the amoimt of noncondensables, the vaporizer temperature and the temperature o f cooling
water. Based on the processing o f 1000 bpd o f a conventional Mid Continent topped
crude oil. Nelson showed that the most economical pressure for vacuum tower operation
is in the range from 30 to 100 mmHg. The steam consumption at 88 mmHg for different
Table 5.1 Vacuum distillation steam (process steam + vacuum jet steam)
consumption for 1000 bpd topped crude (Nelson, 1958)
Vapor line absolute Vaporizer Cooling water Steam consumption
pressure, mm Hg temperature, °C temperature, °C lb/day
88 360 27 58800
88 382 27 23280
The optimum performance depends on the absolute pressure in the flash zone and
the maximum non-cracking temperature at the vacuum heater outlet. Figure 5.5, taken
from Watkins (1979) shows the ratio o f gas oil distilled as a function of the flash zone
The system pressure in this figure is the partial pressure o f the hydrocarbons
instead of the total pressure. In turn, the furnace outlet temperature depends on the
boiling range o f the feed, the fraction vaporized as well as on the feed coking
characteristics. Normally, the maximum flash zone temperatures in the vacuum tower
137
The absolute pressure in the vacuum tower flash zone is controlled between 25
and 100 mmHg. The effective pressure may be reduced to about 10 mmHg by injecting
steam to the furnace and at the bottom of the vacuum tower. The amount o f stripping
steam used is a function o f the boiling range of the feedstock and the fraction vaporized,
and generally in the range o f 10 -50 Ib/bbl. The desired vacuum is maintained by the use
o f steam ejectors and barometric condensers. The number o f ejectors used is determined
by the vacuum needed. For a flash zone pressure o f 25 mmHg, three ejector stages are
120
0% 30% 50% 80%
100
20
Vacuum distillation is widely used to produce catalytic cracking plant feed stocks
o f low carbon content. It is also used to produce lube oil fractions. In practice, the uses of
138
vacuum distillates and residue depend on the type o f crude oil feed, the type o f refinery
and its downstream processing capacities. The major specifications for the most common
• Catalytic cracking feedstocks: Gas oils for catalytic cracking feedstocks require a
strict separation between distillate and residue. The amount o f carbon content in these
oils should be minimal while not sacrificing gas oil recovery. This is necessary to
reduce coke forming on the cracking catalyst. The metals content, particularly
vanadium and nickel, should be strictly limited because they are severe catalyst
poisons.
hydrocrackers can tolerate a slightly higher metal content because the hydrogenation
catalysts contain these metals. However, the amount of carbon and asphalt materials
• Distillate fu el oils: Distillate fuel oils to be used directly are specified by API gravity,
viscosity, metal content and flash point. Atmospheric boiling ranges can be used to
• Vacuum residue: The properties of vacuum residue can be defined in several ways.
remaining in the bottoms stream must be minimized. The residue is blended into
residual fuels. In this operation, one can normally set the volume percent o f either the
whole crude or the topped crude that is to be yielded as vacuum residuum. This is in
effect equivalent to specifying a TBP cut point for vacuum distillates. The goal is to
139
recover the maximum volume o f gas oil, which is free from contaminants by heavier
material.
Table 5.2 shows typical specifications which are used in this study. This is a fuel-
type tower with two distillate products: light vacuum gas oil (LVGO) and heavy vacuum
gas oil (HVGO). In the refining industry, the specification for the LVGO is its flowrate,
and it is usually defined to be one third of the total distillate. In this study, however, the
same flowrate for LVGO cannot be specified in the three complete plants, because the
flowrate o f LVGO varies in a wide range. Therefore, D86 95% temperature is used to
Products from the vacuum tower are cooled down to recover heat before being
sent for further processing. Naturally, one would like to recover as much heat as possible.
However, the final temperatures o f these products should not be too low, otherwise, they
become too viscous to pump. Table 5.3 shows the viscosity o f LVGO, HVGO and
140
vacuum residue at several temperatures. The final temperatures for LVGO, HVGO and
conventional -vacuum design, the Watkins design method is used to obtain an initial
constructed, and the direction o f heat shifting needed for maximum energy efficiency is
Step 1: Begin with the lightest crude to be processed. As suggested previously, the
atmospheric tower does not have any pumparounds at this stage. The vacuum tower,
however, needs to have at least one top pumparound because it does not have a
condenser.
141
Step 2: The simulation is performed next. Usually simulation converges without
difficulty.
Step 4: Transfer heat from the condenser of the atmospheric tower to the top atmospheric
pump-around.
Step 5: If product gaps (difference between 5% ASTM D86 temperature o f the heavier
product and the 95% ASTM D86 temperature of the lighter adjacent product) becomes
smaller than required, the stripping steam flowrate is increased to fix the gap. As long as
the steam added has a lower cost than the energy saved, one can continue shifting loads.
Step 6: If there is heat surplus from the pump-around circuit just added, transfer the heat
to the next pump-around circuit between draws in the same way as in step 4. If not, stop.
Step 7: For the vacuum tower, shift heat from the top pump-around to the low pump-
around. Check the total energy consumption and the product gap. The heat shifting
should stop when either the energy consumption does not decrease or the minimum
E = U + 0 .7 * '^ H ; (5.1)
where, U is the minimum heating utility of the system and H ' is the enthalpy of each
steam used in the column strippers. The weight factor o f 0.7 that multiply the steam
enthalpies is used because the low-pressure steam is cheaper than the fuel gas used as
142
heating utility. In turn, the minimum heating utility is assumed to be provided by a
The presence o f the vacuum distillation affects the optimal heat load distribution
in the atmospheric tower. With the variation o f heat distribution in the vacuum tower, the
pinch location may change. When Step 7 is completed, one should check the heat
demand-supply diagram and see whether the heat distribution in the atmospheric tower is
optimal. If not, one has to adjust the heat distribution and go to Step 5 again.
The pre-flash/pre-fractionation case: A special feature o f the preflash design is that the
heat demand for the crude can be adjusted by changing the temperature o f the preflash
drum. Because the total heat load in the atmospheric tower depends on the duty o f the
Step 3: Increase the temperature o f the preflash drum until the total energy consumption
starts to increase. During this procedure, product gaps are maintained by adjusting the
stripping steams.
The stripping-type atmospheric and vacuum case: Instead o f having several pump
arounds as in the atmospheric tower of the conventional design, the atmospheric tower in
the stripping-type design has several heaters. With the knowledge that the conventional
design starts with no pumparounds, one can propose a stripping-type design to start with
only one heater. However, simulation shows this is not feasible because o f the
143
temperature limits. To overcome this problem, an initial heat distribution among the three
Step 1: Begin with a stripping-type tower with an initial heat distribution. Set an initial
Step 3: Increase the temperature o f the feed until the total energy consumption starts to
increase.
5.7 Results
It is shown in previous work (Ji and Bagajewicz, 2001) that both preflash and
stripping-type designs produce more residue than the conventional design. The higher
feed rate to the vacuum tower has two immediate effects. One is the requirement of a
higher capacity o f the vacuum tower and the other is a higher suction load to the vacuum
producing system. From the viewpoint o f energy consumption, both preflash design and
the stripping-type design have the advantage o f reducing the direct heat demand o f the
crude oil by avoiding overheating o f the light components. However, the corresponding
vacuum system consumes more steam. The trade-off both the two factors is here
The feedstock used in this section is a light crude o f API 36. The property data for
The conventional atmospheric-vacuum distillation design: Basically, there are two rules
to determine if a new pump-around is needed (Step 6). One is the existence o f heat
surplus in the pump-around last added and the other is that the addition o f a new pump
around should reduce the total energy consumption. For the vacuum tower, the two
144
pump-arounds already exist at the beginning o f the procedure. The problem is then to
Table 5.4 shows the resulting energy consumption after the corresponding steps in
the targeting procedure are implemented. The energy consumption decreases constantly
with the addition of a new pump-around. The resulting heat demand-supply diagram is
Table 5.4. The variation o f the energy consumption with the addition of
each pump-around in the case o f the conventional-vacuum design
(Vacuum jet steam consumption not included)
- Energy consumption (MW)
No piunparound (Step 3) 97.29
With PAl (Step 4) 95.33
With PAl and PA2 (Step 6) 86.46
With PAl, PA2 and PA3 (Step 6) 76.94
With VPA2 optimized (Step 7) 75.13
ATmin = 22.2 ° C.
consumption. This is because the temperature o f the heat provided by PAl is still low and
in the heat demand-supply diagram, most of this heat is in the region o f the condenser,
where heat supply is already in surplus. The reason why VPA2 results in a small decrease
is that the duty of the VPA2 is relatively small compared to the duties o f atmospheric
pump-arounds.
145
1.2
Crude
PAl
0.8
I
0.6
Q. Cond VPA2
U < V4
s
^ V3
0.4
V4
Vacuum
PA1-
SW PA3 furnace
0.2 PA2 V4
gas oil
kero
0
100 200 300 400 500
diesel
TEM PERATURE, C
(atmospheric-vacuum plant)
146
Although the heat loads provided by vacuum pump-arounds and products are not
as large as those o f the atmospheric counterparts, they affect the heat distribution in the
intermediate temperature region (PA2 region), it is beneficial for the complete design to
shift more heat from PA2 to PA3. Table 5.5 compares the heat distribution in the
atmospheric tower with the conventional design (Bagajewicz and Ji, 2001).
The complete plant has a lower duty in the middle pump-around (PA2) and a
higher duty in the lower pump-around (PA3). This can be explained as follows. In the
atmospheric conventional distillation only (see Figure 5.7), when the PA3 duty reaches
8.79 MW, the surplus in the PA2 region vanishes. Therefore, further heat shift to PA3
does not reduce the net heat demand and worsens the separation between diesel and gas
oil.
147
1.4
.2
1 PINCH
g 0.8
CRUDE
dlQ.6 COND
0.4
PA2 PA3
0.2
RES
0
0 100 200 300 400
TEMPERATURE. °C
148
In the complete atmospheric-vacuum plant, however, vacuum products and pump
arounds provide new heat sources in the PA2 region (Figure 5.6), which contribute to a
larger heat surplus in the PA2 region and allows more heat to be shifted from PA2 to
PA3. The duty of PA3 increases until the trade-off between the reduced energy
5.6. In this table from the left to right are the conventional design, the preflashing design
and the stripping-type design. The preflash design has the lowest energy consumption,
but its difference with the conventional is small, indicating that the conventional design is
competitive. The difference is because the atmospheric heater in the preflash design has a
Figure 5.8 shows the heat demand-supply diagram for the preflash type crude
distillation. In this figure, the heat demand for the crude without preflashing is shown as
dashed line.
The Stripping-type design consumes 40% more energy than the conventional
design. Both the steam consumption and the minimum heating utility are higher than that
of the conventional design. The larger vacuum steam consumption is incurred by the
larger amount feed to the vacuum tower. The vacuum steam consumption of the
stripping-type design is 48.5% higher than the conventional design. The large increase in
the duty o f the atmospheric tower is due to the requirement o f heat at higher temperature
levels where less recoverable heat is available. The requirement for higher temperature
heat makes a large part o f the heat from the products and the pump-arounds unusable, and
149
1.2
VPA1
V3
V2
PAl Pinch
Cond Crude
VPA1
d
VPA2
V4
0.4
V3
Vacuum
Furnace
0.2
Jsw
gas c
kero
Figure 5.8 Heat demand-supply diagram for preflash type distillation (light crude)
150
Table 5.6. Comparison o f the optimal designs o f three complete plants
Conventional Pre flashing Stripping-type
Steam enthalpy, MW
Atmospheric 15.74 13.08 16.38
Vacuum* 11.55 12.35 17.15
Total steam 27.28 25.43 33.53
Product Gap, °C
Naphtha-kerosene 16.6 16.7 18.8
Kerosene-diesel 0.0 0.0 0.7
Diesel-gas oil -4.0 -4.5 -25.8
LVGO-HVGO -30.4 -29.7 -30.2
Yield,M^/hr
Naphtha 244.39 244.92 244.81
Kerosene 144.76 145.15 141.03
Diesel 71.82 69.95 51.84
Gas oil 124.25 110.40 45.44
LVGO 22.92 34.92 115.09
HVGO 81.43 86.09 92.24
Vacuum residue 105.53 103.58 104.37
Vacuum overhead, kg/hr 171.95 219.04 92.08
HDBR**, MW 195.81 191.58
HU, MW 62.57 62.10 80.48
Energy, MW 81.67 79.90 114.01
151
2
.6
LH
u
.2
UH
HVGO MH
.8 LVGO VPA2
VRES /
SC3
CONDENSER
VPA VH
0.4
G
^Gas oil, 1 0 0/ / 200 SCI 300 400 500
SC2
Diesel VRES
TEMPERATURE. C
152
It is also seen that both the stripping-type design and the preflash design produce
less atmospheric gas oil, but the loss is picked up by the vacuum tower in terms of more
LVGO. The product gaps are similar among the three designs except for the diesel-gas oil
gap in the stripping-type design, where some diesel is not vaporized in the mid heater and
Figure 5.9 shows the heat demand-supply diagram for stripping-type crude
distillation. The addition o f the vacuum tower does not have significant effeci on the
topology of the heat demand-supply diagram. This is because the heat demand takes
place at high temperatures. The heat from vacuum products and vacuum pump-arounds
appears on the left o f these heaters, where the heat can only be used to preheat the crude.
While the heat from the vacuum section can be utilized to a large extent in both the
conventional design and the preflash design, it is almost o f no use in the stripping-type
design. This explains why the energy efficiency for the stripping-type design in a
complete plant is even worse than in the stripping-type atmospheric distillation plant.
As mentioned before, both the preflash design and the stripping-type design aim
at light crudes. In terms o f energy efficiency, the possible gain from both designs is that
the heat demand for the light components can be lowered (Ji and Bagajewicz, 2002a, b).
Because the amount o f light components in a heavy crude is low, it does not make sense
to use the preflash design or the stripping-type design for a heavy crude. Therefore, only
the conventional-vacuum design will be studied for the heavy crude. The heavy crude o f
20.0 API is used in this study. Property data o f this crude are shown in the appendix.
153
The heating utility for a complete conventional plant processing a heavy crude is
51.35 MW. The heat demand-supply diagram and the operation variables for a scheme
with three pump-around circuits are shown in Figure 5.10. Compared with the
conventional design without vacuum distillation (Figure 5.11), the heat deficit in the low
to medium temperature range is improved but still prevails. The heat surplus in the PA2
region can be used to cover the heat deficit in the condenser region. Similar to the
atmospheric design, the total energy consumption is not sensitive to the atmospheric
tower heat distribution. This is an added flexibility for the heat exchanger network
design.
5.9 Conclusion
Rigorous targeting procedures have been developed for three types of complete
crude distillation plants. It has been found that the introduction o f vacuum tower changes
the topologies for both the conventional design and the preflash design, and thereby
changes the heat distribution among the pumparounds. In the stripping-type design,
however, the heat provided by the vacuum products cannot be utilized. The comparison
shows that energy consumption for the preflash-vacuum design is slightly smaller than
small, it should not be concluded that the preflash-vacuum design is better. The energy
targets obtained above are used in Part II to develop a heat exchanger network for a
154
.2
1
VPA2
0.8 VPA1 Crude
0.6
S 0.4
Cond
0.2 Residue
SW PA2
0
0 100 200 300 400 500
TEMPERATURE, 0
155
0.8
0.6 CRUDE
PA3
üo
0.4
Q.
RES
ü
COND
* 0.2
PA1
SW PA2
Figure 5.11 The conventional crude distillation for the heavy crude
156
5.10 References
Company, 1964.
4. Gary, James H., Handwerk, Glenn E., Petroleum Refining; technology and
5. Golden, S., Prevent Pre-flash Drum Foaming, Hydrocarbon Processing, May 1997,
ppl41-153.
6. Harbert, W. D., Preflash saves energy in crude unit. Hydrocarbon Processing, July,
123-125, 1978.
7. Ji, Shuncheng and Bagajewicz, M., Rigorous targeting procedure for the design of
8. Ji, Shuncheng and Bagajewicz, M., On the energy efficiency of stripping-type crude
9. Liebmann, K., Integrated crude oil distillation design, Ph.D. dissertation. University
10. Liebmann, K., and Dhole, V. R., Integrated Crude Distillation Design. Computers &
157
11. Liebmann, K.; Dhole, V. R. and Jobson, M., Integration Design O f A Conventional
12. Maxwell, J.B., Data Book on Hydrocarbon, Princeton, N.J., D. van Nostrand Co.,
1965.
13. Nelson, W. L., Petroleum Refinery Engineering, 4 ^ ed.. New York, 1958.
14. Packie, J.W. Distillation Equipment in the Oil Refining Industry. AIChE
15. Bagajewicz, M. and Soto, J., Rigorous Procedure for the Design of Conventional
Atmospheric Crude Fractionation Units. Part III; Trade-Off Between Complexity And
16. Watkins, R. N., Petroleum Refinery Distillation. G ulf Publishing Company, 1979.
5.11 Appendix
Appendix A
The crude oils used in this paper are summarized in Table 5A.1. The process rate
is 5000 bbl/hr. The TBP data and light ends composition are shown in Table 5A.2 and
Table 5A.3.
158
Table 5A.2 TBP Data (°C)
Vol. % Light Crude Intermediate Heavy Crude
Crude
5 45 94 133
10 82 131 237
30 186 265 344
50 281 380 482
70 382 506 640
90 552 670 N/A
Appendix B
Table 5A.4 lists the operating conditions o f the designs reported by Liebmann
(1996). The designs w ere carried out using ASPEN plus release 8.5-6. Column 1 is the
159
Table 5A.4 Comparison o f Operating conditions in two types of designs
Liebmann-D-2 Base-Liebmann***
Flash zone in, C 343
FZT,C 340
FZP,kpa 261
Heaters:
Heater-12 (UH) 195.2-248.5 C
Heater-23 (MH) 202.5-370 C
Heater-34 (LH) 340.6-370 C
*Reboiler ** Crude 100,000 BBL/day
160
Chapter 6 Design of Crude Distillation Plants with Vacuum
6.1 Overview
In this chapter, the design o f the heat exchanger network is presented. A multi
period heat exchanger network design model is proposed to handle two different crudes —
- a light crude and a heavy crude. This model considers the existence o f a preflash drum,
which is only used for the light crude period. Part of this multi-period model contains a
topological constraint through which all periods share the same heat exchange “matching
6.2 Introduction
In Chapter 5, rigorous targeting procedures have been developed for three types
of complete crude distillation plants. It was found that the introduction o f vacuum tower
changes the topologies for both the conventional design and the preflash design, and
thereby changes the heat distribution among the pumparounds. Comparisons show that
energy consumption for the preflash-vacuum design is slightly smaller than the
compete with the conventional design in both the operating cost and the capital
investment. The energy targets obtained above are used in this part to develop a heat
161
Bagajewicz and Soto (2001a) presented heat exchanger networks for atmospheric
distillation featuring maximum energy efficiency. The problem with such a design is that
it requires extensive splitting o f the crude stream. Such high splitting reduces the
controllability o f the preheat train. In another paper by the same authors (Bagajewicz and
Soto, 2001b), heat exchanger network for the atmospheric distillation using two, three
and four branches, respectively, were studied and showed the total cost is not sensitive to
the number o f branches. This is because the HEN with less number of matches includes
less exchanger shells and therefore less investment cost. It was concluded that when the
branching is reduced, the trade-off between heat utility and capital cost breaks even.
The model proposed in the above paper (Bagajewicz and Soto, 2001b) uses binary
variables to count heat exchangers and these binary variables are shared by both
operation periods. The sharing o f binary variables forces the the same matches for a
given pair o f streams in the same intervals. This paper extends the model to cases where a
match between a given pair can take place in different intervals in different periods. The
heat exchanger network also takes into account the flexibility o f using preflashing units
Although many heat exchanger network (HEN) design techniques are available,
their ability to address rigorous industrial problems with large number of streams and at
the same time controlling branching is limited. For example, Nielsen (1997) noted that
most HEN problems in the literature had less than 10 process streams. Papalexandri
(1998) and his coworkers noted that “industrial applicability o f the proposed methods (for
HEN) has been limited, as important features of industrial HEN can not been efficiently
account for”. For example, too complex heat exchanger networks with many branches
162
and complex piping should be avoided because o f process operability and reliability
concerns.
Besides the main objective o f the maximum heat utilization, the HEN should also
In this regard, Bagajewicz (1998) and Bagajewicz and Ji(2001) showed that operating
conditions (process stream temperatures and flowrates) in a crude distillation plant vary
The problem o f multiperiod HEN has been addressed by several authors. Floudas
embeds all possible structural options for different time periods (Floudas and Grossmarm,
hyperstructure representation of mass and heat exchange alternatives to account for all
The HEN design problem for a complete crude distillation plant discussed in
Chapter 5 has at least 18 process streams (15 hot and 3 cold). Compared with that of
chemical plants, the flexibility issue is much more challenging. Depending on the
difference o f crude oils to be processed, the stream flowrates and temperatures can vary
in such a wide range that a HEN for one crude oil is not even feasible for another crude
oil. With atmospheric distillation, Bagajewicz and Soto (2001) showed the optimal
topology for a light crude HEN departs substantially from that for a heavy crude. In fact,
This chapter addresses the design o f a heat exchanger network for a complete
crude distillation unit that uses a vacuum column for both crudes and a preflash unit for a
163
light crude. A new concept o f “match pattern” is presented and its mathematical
model as in Bagajewicz and Soto (2001) and extends this model to allow the crude to
match the same set of hot streams in different intervals and be considered the same
Figure 6.1 shows the diagram for a complete crude distillation plant. Note that the
preflash drum is bypassed in the heavy crude period. The energy targets are used to
design the heat exchanger network are determined by the procedure proposed in Part I (Ji
sour water
A m
g a so il
steam
ÏÏ
Desalter V Vacuum Residue
Furnace
Vacuum furnace
Heavy crude bypassing
preflash drum
164
The design procedure o f the distillation column starts by assuming a minimum
temperature (HRAT). First, the Watkins design method is used to obtain an initial scheme
the direction o f heat shifting needed for maximum energy efficiency is determined. This
With this procedure and a HRAT value of 22.2 °C, the energy target for light
crude oil is obtained from Figures 2. The same procedure is applied for the heavy crude
case (Figure 6.3) and two sets of conditions can be obtained; product and pump around
flowrates, supply and target temperatures, ^ d energy consumption for the light and
heavy crude. The data that constitute the starting point of the design are shown in Tables
It can be seen from the heat demand-supply diagrams that a heat exchanger
network featuring the minimum utility should be complex. For example, the pinch point
for the light crude (Figure 6.2) is 254 °C. Above the pinch, five hot streams are found in
the region between 254 °C and 310 °C. In order to completely use the heat from all the
five streams, the crude stream has to be split into five branches.
165
2
V PA 1
V3
1 V2
PA1 Pinch
0.8
i Cond VPA1
Crude
0.6
a.
u VPA 2
s V4
0 .4
V3
Vacuum
Furnace
SW
2
gas oil
kero
0
100 200 300 400 500
diesel
TEMPERATURE, C
Figure 6.2 Heat demand-supply diagram for the light crude distillation
166
VPA2
ü 0.8 VPA1 Crude
s 0.6
2 0.4
Cond
0.2 Residue
SW PA2
167
Table 6. I Stream data for light crude distillation with preflashing
168
Table 6. 2 Stream data for heavy crude distillation
169
6.4 Heat exchanger netw ork
As mentioned before, the penalty for limited number o f branches is that the HEN requires
• Each cold stream exchanges heat with a fixed set o f hot streams in the same order in
• HEN should have the ability to handle, different number o f process streams in
different period.
In the case o f light crude processing, a preflash drum is used to flash off light
ends to avoid two-phase flow in the preheat heat exchanger train. The preflash drum is
bypassed in the case o f heavy crude processing. Therefore, the light crude oil period has
The classical transshipment model (Papoulias and Grossmann, 1983) uses one
integer per pair of streams to determine the existence o f a match. This leads to many
streams transferring heat in the same interval and as a result, one cannot control splitting.
Bagajewicz and Soto (2001b) used integers as in the original transshipment model, but
they defined these integers in each interval. To count excahngers, they used a special set
170
Although the model presented by Bagajewicz and Soto (2001) renders good
control o f branching, the model forces the two periods to share the same matches in the
same intervals. Figure 6.4 illustrates a cold stream C; matching with hot streams H, and
H] in two periods. The two topologies are apparently the same, that is, C\ received heat
first from and then from Hi. Only two heat exchangers are needed. However, the
above model can not handle this situation because in the second interval, the constraint of
single match in each interval is violated, even though the cold stream can cascade heat
up.
This paper uses two independent sets o f binary variables to count matches in each
period separately. A constraint is then applied which requires both periods share the same
topology. The topology is expressed in terms of matching patterns, which are defined in
Each cold stream is required to exchange heat with the same hot streams and in
the same order in all operation periods. An algorithm to represent the matching pattern
for each cold stream in each period is needed. Such representation should be unique for
Bagajewicz and Rodera (1998) proposed to count heat exchangers using the
following constraint.
^,.y .r = m ax (y .y ^ - y .^ , 0) (6.1 )
In this constraint, y . j j represents a match between hot stream i and cold stream j
171
HI mcp=l
H2 mcp=l
Cl m C p=l
Period I
© © 1 @
^---------<•------► m cp= l
HI
H2 i w m cp= l
Cl w # 1 m cp= l
Period II
Figure 6.4 The above figures show the same matching sequence between C l and hot
172
Now, consider three hot streams H |, Hz and H3 matching with cold stream J, as
HI H2 H3
Cl
0-0
Figure 6.5 A cold stream matching with three hot streams
Next another parameter PAj j is used to capture the information as to which hot
^^j.T - S ( 6 .2 )
P B j j =mdûi{{PAjj C - (6.3)
*sr-i
where C is a number large enough to ensure that PBj j is non-negative. Further, define a
matching pattern identification number (MPIJ) for a specific sequence o f hot streams as
follows:
(6.4)
173
Under certain choices of a, , MPIj is unique for each matching sequence. Table
6 . 3 shows the details corresponding to an example where J, matches with 3 hot streams
Hu Hz and H3.
T1 T2 T3 T4 T5 T6
0 1 0 0 0 0
0 0 0 1 0 0
0 0 0 0 1 0
PAn.r 0 0 0
P P j\.T
0 0 0
M P Ij
C -l-a„y = C '-2 -a „ ^
A,-hAC2>B
B>A,
B = M o x ( A j , A 2 ) <=> (6.5)
A2 + ( l - A ) n > B
B > A,
For matches between cold stream j and N hot streams, the possible number o f
174
Parent pattern and children patterns
Suppose a matching pattern for a cold stream involves K hot streams in sequence.
If a matching pattern is obtained by removing one or more hot streams without changing
the order o f the remaining hot streams, then the resulting pattern is called a child pattern.
Figure 6.6 shows an example o f a parent matching pattern and its six children patterns.
HI H2 H3
H2 H3
Cl
Cl
o - o — <> 0-0
(a) Parent pattern (b)
HI H3 HI H2
Cl Cl
0-0 0-0
(C) (d)
H2 HI H3
▲
Cl
o Cl
o Cl
o
(c) (0 (g )
175
Thus, for any given cold stream, the matching pattern for one period and has to be
Let MP(N, m) represent the number o f matching patterns for a cold stream to
match with m hot streams which are picked from a hot stream pool o f N hot streams.
Thus,
jV’
MP(N. m )= — (6.6)
m!
CP{K,i) = (6.7)
(6 .8)
y
Consider a cold stream j matching sequentially with N hot streams I,,!?, L , • • In-i ,
N -l
M/>/. (6.9)
1=1 j~ \
Different cases are now considered to infer the values of parameters a, to make
M P I) - M P I j = - ( N - k ) - a „ + ( N - k ) a, (6.10)
where M Pl) is the new pattern. After some manipulation, one obtains:
176
MPIj - M PIj = { m - k ) {a^ -a „ ) ( 6 . 11)
Consider now altering the order of three arbitrary hot streams Ik, I| and Im (k<l<m)
to a new order, say Im, Ik, Ii, keeping the positions of the other matches unchanged. Then
(6 .12)
_ { l —k) üj, + { m - l ) Q/
;
fn -k (6. 13)
In a similar fashion one can obtain the condition for M P Ij = M PI j when 4 hot
(6.14)
/+j + k +l
where i, j, k, I are positive integers that represent differences between the exchanged
positions. In view of the above, to make the value of MPIj unique, parameters a, should
• Be positive.
• Be different.
Table 6. 4 lists parameters a- used in this study. It has been chosen to make a, as
hot stream number plus a decimal number. The purpose of picking such values is to
177
Table 6. 4 Parameter a, in this work
Stream Stream
11 1 .2 1 18 8.05
12 2.03 19 9.13
13 3.27 110 10.33
14 4.53 111 11.16
15 5.04 112 12.24
16 6.07 113 13.41
17 7.31 SI 20
In this paper, most o f the constraints by Bagajewicz and Soto (2001) are used.
However, instead of requiring the same matches interval by interval for different periods,
the matching pattern constraints are used. Following, the sets, variables and constraints
are listed.
Sets:
• Cold streams after splitting: J (including cooling water, which is placed at the last)
Subsets
JB=[Jj, J 4 ] J 3 and J4 are two branches o f J02 (crude after the desalter)
JC=[Js, Jè[ J 5 and ie are two branches o f J03 (crude after the preflash drum)
178
• Hot streams including furnace: H ( furnace appears last)
PA l, PA2, PA3}
Variables:
Let 0(J) be split ratio o f branch J. For each period, the following equations hold:
Y ,d { ja ) = \ (6.15)
JA
Y ,e u b ) = \ (6.16)
JB
=1 (6.17)
JC
Q c ^ ,- e . Qco,^,, (6.18)
R ^ j — R sx-\ ~ H U j ( 6 .20 )
j
^i.T (6 .21)
j
'EQu.r-QCj.r-O ( 6 .22 )
179
U ijj. 3i\d£. j j are positive constants. For simplicity, the same value can be used for all
ij , T.
A cold branch can only match with one hot stream (except heating utility) in each
interval.
(6.25)
i* S
In this constraint, jj. is a binary variable used to count heat exchangers. Note
that K, jj. can be one in a period even if the heat duty in that interval is zero. This stands
for a heat exchanger that is in use in one period and idle in another period.
E S N (6.27)
i.J.T
design above and design below the preflash drum temperature level. The objective
function for the subproblem above the preflash drum is to minimize the heating utility.
The number of exchangers N is initially given a large munber, say 50, and reduced step-
by-step until the utility consumption starts to increase. In addition, in this subproblem
residue heats in the last interval can be non-zero, allowing in this way to cascade down
180
heat to the other subproblem. As no heating utility is required below the preflash drum,
Because the heavy crude bypasses the preflash drum, there is one less cold
streams in the heavy crude period than in the light crude period. In order to make the
number o f cold streams the same, the heavy crude is regarded as going through a dummy
preflash drum. This dummy preflash drum has neither temperature constraint nor vapor
effluent. By doing so, the heavy crude (after the desalter and before the furnace) is
divided into two segments: before and after the preflash drum (Figure 6.7).
Light crude
Heavy crude
TEM PERATURE, C
Figure 6.7 The starting and targeting temperatures o f colds streams in two periods
181
Using the original intervals, heating utility would be required below the preflash
drum for the heavy crude. Table 6 . 5 shows the interval heat balances. Although the heat
balance for interval T , is positive, cold stream C4 has to split into at least 3 branches in
Table 6 . 5 Heat balance for the heavy crude in the first interval below the preflash
drum
Interval T1 Heat cascaded from design above the
MW preflash drum, MW
Cl 0 .0 0 0 .0 0
C4 24.20 0 .0 0
11 0 .0 0 0 .0 0
12 0 .0 0 0 .0 0
13 0 .0 0 0 .0 0
14 6.28 0 .0 0
15 1.79 1.79
16 1.05 0 .0 0
17 4.56 2.57
18 1.51 0.79
19 7.31 0.03
110 6.03 3.40
111 3.26 0.03
112 3.56 1.38
113 15.01 0 .0 0
182
Table 6. 6 compares the targeted furnace duty and the actual duty. The duties for
both the light crude and the heavy crude increase as expected. There are 8 exchangers
(including the furnace) above the preflash drum and 17 exchangers below, totaling 31
exchangers. In the light crude period, hot streams H I, H6, H7, H8 and HI 1 are not used.
Some o f these streams are hot enough to heat the crude above the pinch.
However, they are not selected before in the same temperature interval there are
other hot streams with larger heat capacities. As the cold streams are limited to split into
two branches, the model picks two hot streams with the largest heat capacities. These
streams cannot be used below the desalter because the heat from the condenser is enough
to heat the crude. The situation changes with the heavy crude, where the aforementioned
hot streams are used to heat the crude below the desalter, because the condenser heat is
not enough (see Figure 6.3). The different heat supply scenarios below the desalter
increase the complexity o f the heat exchanger network. As shown in Figure 6.10, nine
exchangers are required to heat the heavy cmde, while only one is used for the light crude
(Figure 6.10). The situation is similar to that found by Bagajewicz and Jose (2001) with
183
6.8 Conclusion
The paper addresses the design o f a heat exchanger network for complete crude
distillation plants. A multi-period heat exchanger network design model was proposed to
handle two radically different crudes. In order to reduce the complexity o f the heat
exchanger network, an assumption is made that crude streams can be split to no more
than two branches. This model also takes into account the flexibility o f using preflash
drum for the light crude only. Part o f this multi-period model contains a topological
constraint through which all periods share the same heat exchange “matching pattern” but
10.33
247 C 185 0
H4
306 C 16.45
o 247 0
HS
313C
o 8.77 185 0
H8
313 C
o 306 0
HIO
4.29
313 C 234 0
H12
o
H13
371 C 9.70
o 247 0
61.70
256 0 207 0 194 0
383 C 257 C
CXX) 0 .6 6 163 0
02
o
270 C 244 0 0.34
11.56 o o
4130 349 0
03
Figure 6.8 Heat exchanger network above the pieflash drum for the light crude
184
3.98
257 C 226 C
H4
313C 1.14 279 C
H5
279 C 279 C
H8
313 C 1.30 289 C
HIO
2.51
313 C 257 C
H12
377 C 46.40 226 C
H13
36.17
2 60C 252C 233 C
363 C 270 C 0.245 203 C
0% H D C2
316C 0.755
22.06
400 C 354 C
Not used
Figure 6.9 Heat exchanger network above the preflash drum for the heavy
crude
185
8.66
145 C
H2 43 C
7.78 9.96
177 C 145 C
H3 !0 4 C
232 C
12
H9 93 C
6.56
306 C 184C 0.38
HIO 177 C
3.80
1.67
247 C 198 C
H13 177 C
38.83
I0 4 C
21 C
Cl
Figure 6.10 Heat exchanger network below the preflash drum for the light
crude
( Coolers for H I, H6, H7, H8, HI 1 and H12 are not shown, idle exchangers on
186
8.77
138 C 43 C
12.83
119 C 43 C
H2
7.33
160 C 104C
H3
6.34
177C
226 C
H4
8.73
1.26
289 C 258 C
H7
226 C 93 C
H9
7.76
226 c 108 C
H Il
5.31
H12
257 C 139 C
i;
56 C 0.42
138 C 117C 86 C
Cl
21 C
0.58
138 C 134 C 58 C
Figure 6.11 Heat exchanger network for heavy crude (C l) below the desalter
H3
9.76 9.98
226 C 160C 93 Ç
H9
6.03
289 C 177 C
HIO
15.04
226 C 177C
H13
203 C
0 56 ,38 c
C4
203 C
0.44
Figure 6.12 Heat exchanger network for heavy crude (C4) below the preflash
drum
187
6.9 N om enclature
H- = enthalpy o f steam i
I = hot stream
J =cold stream
PAj j. = parameter representing a match between a cold stream j and hot stream i in
interval T
Q ijx ~ heat transfer between hot stream i and cold stream j in interval T
Qsj.t~ heat transfer between hot utility S and cold stream j in interval T
T —interval
188
U = minimum heating utility
y, y r - binary variable for heat transfer bet^veen stream i and cold stream j in interval T
6.10 References
1. Andrecovich, M., and Westerberg, A., A Simple Synthesis Method Based on Utility
(1985).
Atmospheric Crude Fractionation Units Part II: Heat Exchanger Networks. Industrial
4. Bagajewicz M. and Soto. J., Rigorous Procedure for the Design of Conventional
Atmospheric Crude Fractionation Units. Part III: Trade-Off between Complexity and
189
7. Bagajewicz M. and H. Rodera. A New MILP M odel For Heat/Mass Exchanger
(1998).
(1987).
9. Floudas, C. and Grossmann, I., Synthesis o f flexible heat exchanger networks for
10. Golden, S., Prevent Pre-flash Drum Foaming, Hydrocarbon Processing, May 1997,
ppl41-153.
12. Ji, S. and Bagajewicz M., Rigorous targeting procedure for the design o f crude
13. Liebmann, K.; Dhole, V. R. Integrated Crude Distillation Design. Comput. Chem.
14. Liebmann, K.; Dhole, V. R.; Jobson, M. Integration Design O f A Conventional Crude
15. Nielsen, J., et al. Retrofit and optimization o f industrial heat exchanger networks: a
(1997).
190
16. Papalexandri K, et al. Heat integration aspects in a crude preheat refining section.
17. Papalexandri, K. and Pistikopoulos, E., a multiperiod MINLP model for improving
the flexibility o f heat exchanger networks. Computers and Chemical Engineering, 17,
8111-116(1993).
18. Papalexandri, K. and Pistikopoulos, E., a multiperiod MINLP model for the synthesis
of flexible heat and mass exchange networks. Computers and Chemical Engineering,
20. Sharma, R.; Jindal, A.; Mandawala, D.; Jana, S. K. Design/Retrofit targets o f Pump-
around Refluxes for Better Energy Integration o f a Crude Distillation Column. Ind.
21. Soto, J., Multipurpose heat exchanger network for maximum energy efficiency of
22. Stichlmair, J.G, and Fair J. R, Distillation Principles and Practices, Wiley-VCH, New
York, 1998.p76.
23. Terranova, B., and Westerberg, A., Temperature-Heat Diagrams for Complex
1374-1379, (1989).
191
Chapter 7 New Flowsheets for Crude Distillation
7.1 Introduction
distillation design. Rigorous procedures were proposed for the design of both the
atmospheric plant and the complete plant. The procedures take into account the flexibility
o f processing different crudes year around. In addition, the role o f light components in
crude fractionation known as the carrier effect or the carrier design was first given a
rigorous definition. In view of this, two typical design alternatives, the preflashing design
and the stripping-type design were analyzed focusing on energy efficiency and compared
design model (Chapter 6) was proposed to handle two radically different crudes — a light
crude and a heavy crude. Part o f this multi-period model contains a topological constraint
through which all periods share the same heat exchange “matching pattern” but not
This chapter explores possible new flowsheets for future study. Two tools are
recovery scenario. Figure 7.1 shows the diagram for the conventional atmospheric crude
192
distillation. Efforts on réduction o f energy consumption should be made in the following
two directions:
0.9 PA1
0.8 Crude
0.7
I
CL
0.6
0.5
Cond
PA1
U 0.4
s PA3
0.3
0.2 PA2
<ero Res
0.1
GO
0 100 200 300 400 500
TEMPERATURE, C
Figure 7.1 Atmospheric Distillation (light crude oil, ATmin - 22.2 °C)
The properties o f the crude can be found in Tables 2.1, 2.2 and 2.3.
Cond: condnser. Kero: kerosene. GO: gas oil. Res: residue. PA: pump-around.
The first case was discussed in Chapter 3. In the second case, one wants to vary
the heating pattern o f the crude, as in the case of the prefash design. In trying to obtain an
alternative, focus is shifted to the utilization of heat from the condenser. The major hurdle
preventing the use of the condenser heat is its low temperature level. The most
193
convenient way to overcome this hurdle is to shift heat from condenser to the pump-
arounds (Bagajewicz, 1998, Bagajewicz and Ji, 2001). However, this effort is limited by
material and heat balance. Specifically, as heat is shifted, the liquid flowrate in the tray
above the pump-around is reduced. When this reaches zero, the maximum transferable
heat is also reached. Now ways to break this limit are explored.
One way to shift the condenser heat to higher temperatures is by using a heat
C ru d e
HE
High Temperature
H eat out (Q ^)
(T h)
Low Temperature
H eat in (Q^)
(JO
C ondenser
This alternative does not alter the operating conditions of the column, but is
expensive. Table 7. 1 shows the relative cost for shifting heat from the condenser and
194
PAl to the targeted region (Figure 7.3). Note that the heat is transferred through the
pinch.
P in ch
Crude
Ü 0.7
Cond
- 0.5
O 0.4
Figure 7.3 Shifting heat to a high temperature region using heat pump
The properties of the crude can be found in Tables 2.1, 2.2 and 2.3.
The theoretical minimal work (ideal work) needed for transferring the heat can be
195
(7.1)
^H
T;, is the target temperature at which the heat pump discharges heat.
(7.2)
where
through the pinch. The temperature range o f the heat available is 144.1-170.6 °C (417.3-
443.7 °K) and the targeted temperature range is 201.5-254.4 °C (474.7-527.6 °K).
Assume two heat pumps are used, each transferring 6.625 MW heat from the condenser.
The first heat pump operates between 144.1 C and 227.9 C, and the second heat pump
operates between 157.4 C and 254.4 C. The heat pumps have an efficiency of 60% (Table
7.1). Based on the prices provided by Douglas (1988, fuel S4.00/MMbtu, electricity
196
Table 7. 1 Relative operation cost for shifting heat to the targeted region
Distillation sequencing is another useful tool for developing new design schemes.
Figure 7.4. The bottom parts o f columns I, 2 and 3 correspond to the side strippers in the
conventional design (Figure 7.5). The stripping-type design is a direct sequence (Figures
197
I I Condenser
1-------------- ► Naphtha
^----- Steam
----------- ► Kerosene
- Steam
Diesel
Steam
Crude ^ Gas oil
Residue
water
<------- '------- ►na
naphtha
PAl
steam Stripper 3
PA2
c : kerosene
steam Stripper 2
PA3
c : diesel
198
Naphtha
Crude
-► Kerosene
J
ÜH Diesel
^-0 -
MH Gas oil
^— 0 -
LH ■* ----- Steam
Residue
Water
Prch cater
Naphtha
Crude oil ^ Kerosene
Rectifier
UH
Diesel
Rectifier 2
MH
Gas oi
Rectifier 3
LH
Steam
199
Preflashing and Prefractionation designs are other sequences (Figure 7.8, 9). Note
that in the preflash design the sloppy separation o f the first flash produces a top vapor
that needs further separation. In this regard, this scheme departs from the traditional sharp
sequence proposed as framework for the analysis. The complete list of possible
» Kerosene
PA2
* Naphtha
Crude Diesel
PA3
Gas oil
Residue
♦ N a p h th a
PAl
Kerosene
Crude
Diesel
♦ Gas oil
Residue
200
N K
K D
-► — Stripping-type design
D G R
G R
R
K
N
D
G
K K
N D R
D
K G G
D K
R Prefractionation design
G D
N
K
D N
Conventional design
K
In screening the potential sequence, one has to take account of the following
constraints
201
3. Non-sharp split between distillates
These constraints can be used to eliminate some sequence. For example, the first
and the second constraints exclude the stripping-type design. A previous study (Ji and
Bagajewicz, 2001) shows that early separation of stripping components increases the
yield o f the residue, which is not a desired effect. The stripping components include all
compounds in naphtha and kerosene as well as some components in diesel. Therefore, the
residue has to be separated first in order to maintain the stripping effect as in the bottom
branch which eventually leads to the conventional design but renders others too. The
using atmospheric pressure. The current crude vacuum distillation only uses one column
operating at about 10 mmHg. Because the vacuum jet steam consumption increases with
the decrease o f the distillation pressure, it is natural to think about an alternative of using
two vacuum towers, one operating at low vacuum, say 100 mmHg, and the other
operating at 10 mmHg (Figure 7.11). If a significant amount o f vacuum gas oils can be at
the low-vacuum tower, the feed to the second vacuum tower can be reduced and
accordingly the total vacuum jet steam will be reduced. The disadvantage is the extra
investment cost o f the low-vacuum tower. Hopefully, the cost is not high. As the vapor
volume is inversely proportional to the operating pressure, the diameter of the low-
202
To Ejectors To Ejectors
Heavy
X
Topped
crude
»
n
(-—\
diesel
VLGO-I a X
-► VLGO-II
-► VHGO
203
7.5 References
4. Ji Shuncheng and Bagajewicz, M., Rigorous targeting procedure for the design of
5. Liebmann, K., and Dhole, V. R., Integrated Crude Distillation Design. Computers &
204